FLIR Photon, A320, A325 Manual Book

IR Automation Guidebook:
Temperature Monitoring and Control with IR Cameras
$29.95
IR Automation Guidebook:
Temperature Monitoring and Control with IR Cameras
ii
Published by FLIR Systems Incorporated This booklet may not be reproduced in any form without the permission in writing from FLIR Systems Incorporated.
www.goinfrared.com • 1 800 GO-INFRA © Copyright 2008. All rights reserved.
USA, Canada and Latin America
FLIR Systems Incorporated
America’s Main Oce, USA Boston, MA 1-800-GO-INFRA (464-6372) or 1-978-901-8000
Europe, Middle East, Asia and Africa
FLIR Systems
International Main Oce, Sweden Tel: +32 3 287 87 10
iii
Contents
Preface iv
Chapter 1
Typical Monitoring and Control Applications 1
Chapter 2
Remote IR Monitoring 5
Chapter 3
Temperature Measurement for Automated Processes 17
Chapter 4
Combining Machine Vision and Temperature Measurement 25
Chapter 5
Real-Time Control Issues 32
Appendix A
Glossary 40
Appendix B
Thermographic Measurement Techniques 43
Appendix C
History and Theory of Infrared Technology 45
Appendix D
Command Syntax Examples for A320 Resource Socket Services 58
Appendix E
Quick Summary of FLIR IR Cameras
Inside Back Cover
iv
Preface
Manufacturing and process engineers are under constant pressure to make production systems and processes more
ecient and less costly. Frequently, their
solutions use automation techniques to
improve throughput and product quality. Automated IR (infrared) radiation imaging oers the potential for improving a host of industrial production applications,
including process monitoring and
control, quality assurance, asset management, and machine condition
monitoring.
This handbook is intended to help those
considering the creation or improvement
of production automation or monitoring
systems with IR cameras. Numerous
application examples will be presented
with explanations of how these IR vision
systems can best be implemented.
Some of the major topics that will be
covered include:
Integration of IR cameras into automation systems
Data communications interfaces Command and control of
thermographic cameras Principles of thermographic
measurements Interfacing with a PC or PLC controller Standard software packages for IR
camera systems
These complex matters require attention
to many details; therefore, this handbook cannot answer every question a system designer will have about the use of
IR cameras in automated systems. It
is meant to serve only as a roadmap
through the major issues that must be
faced in IR vision system design.
1
Typical Monitoring and
Control Applications
Typical Monitoring and
Control Applications
Temperature Measurements with IR Cameras
Infrared (IR) radiation is not detectable by the human eye, but an IR camera can convert it into a visual image that depicts thermal variations across an object or scene. IR covers a portion of
the electromagnetic spectrum from
approximately 900 to 14,000 nanometers (0.9–14 µm). IR is emitted by all objects at temperatures above absolute zero, and
the amount of radiation increases with
temperature. A properly calibrated IR
camera can capture thermographic images
of target objects and can provide accurate non-contact temperature measurements of those objects. These quantitative measurements can be used in a variety of
monitoring and control applications.
In contrast, other types of IR imagers provide only relative temperature
dierences across an object or scene.
Hence, they are used to make qualitative assessments of the target objects,
primarily in monitoring applications where thermal images are interpreted
based on temperature contrast. One
example is to identify image areas that
correlate to physical anomalies, such as construction or sub-surface details, liquid levels, etc.
In some cases, an IR camera is justiably
referred to as a smart sensor. In these
cases the IR camera has built-in logic
and analytics that allows the comparison
of measured temperatures with user-
supplied temperature data. It also has a
digital I/O interface so that a dierential
temperature can be used for alarm and
control functions. In addition, a smart
IR camera is a calibrated thermographic
instrument capable of accurate non-
contact temperature measurements.
IR cameras with these capabilities operate much like other types of smart
temperature sensors. They have fast, high-resolution A/D (Analog to Digital) converters that sample incoming data, pass it through a calibration function, and provide temperature readouts. They may also have other communication interfaces that provide an output stream of analog
or digital data. This allows thermographic images and temperature data to be transmitted to remote locations for process monitoring and control.
Generally, smart IR cameras are used in quantitative applications that
require accurate measurements of the temperature dierence between a target object and its surroundings. Since temperature changes in most processes
are relatively slow, the near-real-time data
communications of smart IR cameras are adequate for many process control loops
and machine vision systems.
Automation Applications
Typical automated applications using IR cameras for process temperature monitoring and control include:
Continuous casting, extrusion, and roll
forming Discrete parts manufacturing Production where contact temperature
measurements pose problems Inspection and quality control Packaging production and operations
Chapter 1
2
Chapter 1
Environmental, machine, and safety
monitoring Temperature monitoring as a proxy for
other variables
The examples below demonstrate a wide
range of applications that can be served
with IR cameras. Potential applications are limited only by the imagination of the system designer.
Plywood Mill Machine Monitoring
Problem: Steam from open vats of hot water obscures the machinery operator’s view of the logs as they are maneuvered for proper alignment in the log vat.
Solution: An IR camera can present an
image to the operator that makes the
cloud of steam virtually transparent,
thereby allowing logs to be properly
aligned in the log vat. This example of a qualitative application is illustrated in Figure 1.
Production Testing of Car Seat Heaters
Problem: Using contact temperature
sensors to assure proper operation of optional car seat heaters slows down production and is inaccurate if sensors are not properly placed.
Solution: An IR camera can detect
thermal radiation from the heater
elements inside the seats and provide an accurate non-contact temperature
measurement.
This quantitative measurement can be
made with a camera that is permanently
mounted on a xture that is swung
into measurement position when the car reaches a designated point
on the assembly line. A monitor near that position provides an image with a temperature scale that reveals the
temperature of the car seat heater
elements, as shown in Figure 2.
The Problem
• Operatorscannotseethroughthesteam cloudcausedbycondensationincoolerair temperatures.
The Solution
• IRoersanotherpairof“eyes”tosee throughthesteamintothelogvatfor properlogalignment.
Figure 1. Plywood mill application
3
Typical Monitoring and Control Applications
Packaging Operations
Problem: On a high-speed packaging line, ecient methods for non­destructive testing of a glued box seal are scarce, and most tend to be very cumbersome. In addition, the glue
application method has a good deal of
variability that must be monitored and
recorded with statistical quality control routines.
Solution: Since the glue is heated prior
to application, its temperature and
The Problem
• Optionalfeaturesinvehiclescannotbe inspectedwithoutsometypeofcontact.
• Thisslowsdownproduction.
• 100%inspectionistedious.
The Solution
• AnIRcameracanbepermanentlymountedto inspecttheseitems.
• AnIRcameracanbeusedinanon-contact method.
Figure 2. Production testing of car seat heater elements
The Problem
• Detectincorrectlysealedboxes.
• Removefailedunitsfromtheline.
• Generateanalarmiftoomanyboxesfail.
• Logstatisticaldataofpass/fail.
The Solution
• Captureathermalimageofthebox.
• Detectpresenceofgluespots.
• Pass/failoneachbox.
• Logstatistics.
Figure 3. Machine vision box seal quality control
4
Chapter 1
locations on the box lid can be monitored
with an IR camera. Moreover, the image can be digitized in a way that allows this
information to be stored in a statistical quality control database for trend analysis and equipment monitoring as shown in
Figure 3.
This is an example of using dierential temperature as a proxy for another
variable. In this case, temperature
replaces mechanical methods of inspection/testing.
Summary
The automation examples presented
in this chapter have barely scratched
the surface of the application space
that smart IR cameras can serve. In the following chapters, more detailed
examples will be presented along with practical information on the implementation of automated systems
that exploit the advantages of IR cameras. These chapters are organized according
to the major types of applications that typically use IR cameras:
Remote thermographic monitoring
Non-contact temperature
measurement for automated processes
Combining IR machine vision with
temperature measurement
Real-time control and monitoring –
issues and answers
5
Remote IR MonitoringChapter 
Remote IR Monitoring
Overview
Infrared radiation is emitted by all objects
at temperatures above absolute zero
and is detectable by IR cameras. Since
these cameras have various means of
communicating thermographic images
and temperatures to remote locations,
they are ideal for remote and unattended
monitoring. Moreover, smart IR cameras (those with built-in logic, analytics, and data communications), can compare
the temperatures obtained from their
thermographic images with user-dened
settings. This allows the camera to output a digital signal for alarm and
control purposes, while also providing live images.
IR Camera Operation
IR camera construction is similar to
a digital video camera. The main
components are a lens that focuses IR
onto a detector, plus electronics and
software for processing and displaying thermographic images and temperatures
on an LCD or CRT monitor (Figure 1). Instead of a charge coupled device that video and digital still cameras use,
the IR camera detector is a focal plane
array (FPA) of micrometer size pixels made of various materials sensitive to
IR wavelengths. FPA resolution ranges from about 80×80 pixels up to 1024×1024 pixels. In some IR cameras, the video
processing electronics include the logic and analytical functions mentioned
earlier. Camera rmware allows the user to focus on a specic area of the FPA or use the entire detector area for calculating minimum, maximum, and average temperatures. Typically,
temperature measurement precision is ±°C or better.
The camera lens and distance to the
target object results in a eld of view (FOV) that determines the spot size covered by each pixel. The pixel’s analog
output represents the intensity of heat
energy received from the spot it covers on the target object. In FLIR IR cameras, the A/D converters that digitize the pixel output have resolutions that range from 8 bits (28 or 0–255 pixels) up to 14 bits (214 or 0–16383 pixels). The thermographic
image seen on the monitor screen is the result of a microprocessor mapping
these pixel output values to a color or gray scale scheme representing relative temperatures. In addition, radiometric
information associated with the heat energy impinging on a pixel is stored for use in calculating the precise
temperature of the spot covered by
that pixel.
IR In
Optics
NIR
MWIR
LWIR
Video Processing Electronics
Detector Cooling
Digitization
User Interface
User Control
Video Output
Digital Output
Synchronization In/Out
System Status
Figure 1. Simplied block diagram of an IR camera
6
Chapter 
Hence, IR cameras with these capabilities
operate much like other types of smart temperature sensors. Their calibrated
outputs can be accessed via one or more
communication interfaces and monitored
at a remote location. Images saved from
these cameras are fully radiometric1 and
can be analyzed o-line with standard software packages, such as those available from FLIR.
Important Criteria in Remote Monitoring Systems
When considering an IR camera for a
remote monitoring system, some of the important variables to consider are:
Spot size – the smallest feature in a
scene that can be measured
FOV (Field of View) – the area that the
camera sees
Working distance – distance from the
front of the camera lens to the nearest target object
Depth of eld – the maximum depth of
a scene that stays in focus
Resolution – the number of pixels and • size of the sensor’s active area
NETD (Noise Equivalent Temperature • Dierence) – the lowest level of heat
energy that can be measured
Spectral sensitivity – portion of
the IR spectrum that the camera is
sensitive to Temperature measurement range,
precision, and repeatability – a function of overall camera design
1 Radiometry is a measure of how much energy is
radiating from an object, as opposed to thermography,
which is a measure of how hot an object is; the two are related but not the same.
Another fundamental consideration is which portion of a camera’s FOV
contains the critical information required for monitoring purposes. The
objects within the FOV must provide an
accurate indication of the situation being
monitored, based on the temperature
of those objects. Depending on the
situation, the target objects may need
to be in the same position consistently
within the camera’s FOV. Other application variables related to the
monitored scene include:
Emissivity of the target objects• Reected temperatures within the FOV• Atmospheric temperature and
humidity
These topics will be covered in more
detail in a subsequent chapter.
Remote Asset Monitoring
One type of application where IR cameras are very useful is in remote monitoring of property, inventory, and other assets to help prevent loss and improve safety. Frequently, this involves storage facilities,
such as warehouses or open areas for bulk materials. The following example
can serve as a general model for setting
up an IR camera monitoring system for this type of application.
Hazardous Waste Storage Monitoring. In this application barrels of chemical waste
products are stored in a covered facility,
but one in which they cannot be totally
protected from moisture. Thus, there is
the possibility of leaks or barrel contents becoming contaminated by air and
moisture, causing a rise in temperature due to a chemical reaction. Ultimately, there is a risk of re, or even an explosion.
7
Remote IR Monitoring
While visible light cameras might be used in such an application, there often is a line-of-sight problem where many of the barrels cannot be seen, even with
multiple cameras positioned throughout
the storage area. In addition, smoke or ames would have to be present before a visible light camera could detect a
problem. This might be too late for
preventative measures to be taken. In contrast, stand-alone IR cameras
monitoring the facility can detect a
temperature rise within their FOV before re occurs (Figures 2a and 2b).
Depending on the camera manufacturer, several monitoring options are available. For instance, the FLIR A320 camera allows a threshold temperature value to
be set internally for alarm purposes. In
addition, the camera’s logic and clock functions can be congured so that a rise
in temperature must be maintained for a certain period of time before an alarm is sent. This allows the system to ignore a
temporary temperature rise in a camera’s FOV caused by a forklift entering the area
to add or remove barrels. Furthermore,
a hysteresis function can also be used to
prevent an alarm from turning o until
the detected temperature falls well below
the setpoint (Figure 3).
Cameras with a digital I/O interface typically provide an OFF/ON type of
output for alarm purposes. The digital
I/O output is either o or on; when on, it is typically a DC voltage or current. For example, the digital I/O output from a FLIR A320 camera is 10–30VDC for loads of 100mA or less. Typically, the digital I/O output is sent to a PLC (Programble Logic Controller) that controls the portion
of an alarm system associated with the monitored area.
A good way to set up the alarm system is to have all cameras congured so they have a high level digital output when the
temperature is below the alarm condition
that holds a PLC in its non-alarm state.
When the alarm setpoint temperature is
detected, the camera’s digital I/O output goes low (typically zero volts) after an appropriate time delay, causing the PLC
Figure 2a. IR image of a hazardous waste storage area showing two spot temperature readings (26.4°F and 16.8°F) that are in the safe range, plus one reading (98.8°F) that is abnormally high.
Figure 2b. A subsequent image of the same area shows that the abnormal reading in 2a has increased further, causing an alarm to go o.
8
Chapter 
to go into its alarm state. This creates a
fail-safe system. If power to the camera is lost, then there is no high level output to the PLC, which treats that event just as if a temperature had reached the setpoint,
thereby causing an alarm. This alerts
personnel that they have either lost the
monitoring function or there is indeed a temperature rise.
Image monitoring. Receiving a warning
based on temperature measurements is
very useful, but the real power of IR­based asset monitoring is in the camera’s
image processing capabilities. Control
room personnel can get live images from IR cameras that visible light cameras
and other temperature detectors
cannot provide. Again, cameras vary by manufacturer, but the most versatile ones oer a variety of data communication
formats for sending thermographic
images to remote locations. Increasingly, web-enabled cameras are used to allow
monitoring from any location where a PC
is available.
Figure 4 illustrates a system using the FLIR A320’s Ethernet and TCP/IP
communication protocols in conjunction with its alarm setpoint capabilities. The Ethernet portion of the system allows
cable runs of up to 100 meters in length.
By communicating a digital alarm directly
to the PLC, it can immediately activate a visual and/or audible alarm. The visual
alarm can appear on an annunciator panel telling the operator where the alarm originated; the operator then goes
to the PC to look at live image(s) of that
location. Images and temperature data can be stored for future reference and analysis.
A320 cameras can also be congured to
automatically send temperature data
and images to a PC via e-mail (SMTP) or FTP protocol whenever the temperature setpoint is reached, thereby creating a record for subsequent review.
Time
Temperature
Threshold T e mperature (Warning On)
Wa rning O T e mperature
Deadband
O Te mp = On Te mp – Deadband
Hysteresis
Also known as deadband
• Can be thought of as another threshold setting – where the smart sensor resets the alarm that was generated when the original setpoint was compromised
Used to prevent signal “chatter
Figure 3. Hysteresis is an important signal processing characteristic of smart IR cameras, which makes monitoring and control functions much more eective.
9
Remote IR Monitoring
In conjunction with a host controller
running FLIR’s IR MONITOR (or other suitable software), temperature data can be captured for trend analysis. The A320
can also supply a digital compression
of the camera’s analog video signal, which can be sent as MPEG-4 streaming digital video over an Ethernet link to a PC monitor. IR MONITOR can be used to set up temperature measurements, image capture, and camera display functions.
This application allows the PC to display up to nine camera images at a time and switch between additional camera
groups as needed. The FLIR IP CONFIG
software can be used to set up each
camera’s IP address.
After the cameras are congured, the
PC used for monitoring does not need to remain on the network continually. By using the FTP and SMTP protocols
within the camera, the user can receive radiometric images upon alarm events or on a time based schedule. Also, any
available PC with a web browser can be used to access the cameras web server for live video and basic control. This web
interface is password protected.
Most IR cameras have an analog video output in a PAL or NTSC format. Therefore, another image monitoring
possibility is to use a TV monitor to
display thermographic video. A single
control room monitor can be used with
a switch to view live images from each
camera sequentially. When the cameras
are properly congured, control room personnel can view scaled temperature readings for any point or area (minimum, maximum, and average) in that image. (See color scales in the screen capture images depicted in Figure 2.) Not only
will the operator know when there
is excessive heat, he or she can see
where it is.
Another example of the innovative functions available in camera rmware
or external software is a feature called
Figure 4. An example of one type of system conguration for remote IR camera monitoring. The system uses a digital alarm output for annunciating an over-temperature condition and transmits streaming MPEG-4 compressed video that allows the scene to be viewed on a PC monitor.
Use Digital Out on each camera to ALARM on AREA MAX
Use the cameras web interface to congure multiple cameras. Set up one AREA in each camera.
10
Chapter 
image masking. This enables the user
to pre-select specic areas of interest
for analysis of temperature data. This
is illustrated in Figure 5, which shows
continuous monitoring of substation hotspots that indicate problem areas.
A similar type of pattern recognition
software can be used for automated inspection in metal soldering and welding and in laser welding of plastic parts. IR cameras can see heat
conducting through the nished parts
to check the temperature of the areas where parts are joined together against
a stored value. In addition, the software
can learn a weld path to make sure this
path is correct, which is accomplished by programming the specic pixels in
an image to be used by the software for
this purpose. Alternatively, the program
developer can save an image of a “perfect” part and then have the software look for minimum, maximum, or delta values that tells the equipment operator
if a part passes inspection. The car seat
heater inspection described in Chapter 1 can be an example of this, and the same
principle is used in the inspection of car window heater elements by applying power to them and looking at their thermographic image.
Power over Ethernet. It should be noted
that a camera with Ethernet connectivity can be powered from a variety of sources, depending on its design. Typically, a
connection for an external DC supply is
used, or where available, the camera is powered via PoE (Power over Ethernet).
PoE uses a power supply connected to the network with spare signal leads not
Figure 5. Masking functionality of the FLIR A320 IR camera, which is also available in some third party software programs.
11
Remote IR Monitoring
otherwise used in 10/100baseT Ethernet systems. Various PoE congurations are possible. Figure 6 depicts one in which
the power source is located at one end
of the network. (Gigabit Ethernet uses all available data pairs, so PoE is not possible with these systems.)
PoE eliminates the need for a separate power source and conduit run for each camera on the network. The only additional cost is for some minor electrical hardware associated with PoE.
Many applications encompass areas that exceed the maximum Ethernet cable
run of 100m. In those cases, there are wireless and beroptic converter options that provide o-the-shelf solutions for communicating over much greater
distances. These are frequently used in the bulk material storage applications described below.
Additional Asset Monitoring Situations
Bulk Material Storage. Many bulk materials are stored in open yards where air and moisture can help promote decomposition and other exothermic reactions that raise the temperature of the pile. This brings with it the threat
of re, direct monetary loss, and safety issues for personnel. In addition, there
is the risk of consequential damages
caused by res, including loss of nearby property, water damage resulting from re-ghting, and production shutdowns.
Materials that are especially prone to spontaneous combustion include organic
wastes (compost, etc.), scrap paper for recycling, wood, coal, and various inorganic chemicals, such as cement and chlorine hydrates. Even in the absence of spontaneous combustion, many bulk materials like plastics pose a re hazard due to sparks or other external
ignition sources.
1
5
Spare Pair
Signal Pair
Signal Pair
4
2
TX
+48V
RX
DC/DC
Converter
3
6
1
5
4
2
3
6
RX TX
5
Spare Pair
4
5
4
Power Sourcing
Equipment (PSE)
Powered
Device (PD)
Figure 6. Schematic depicting spare-pair PoE delivery using the endpoint PSE arrangement.
12
Chapter 
In most cases, prevention is less costly than a cure, and the best prevention is
continuous monitoring of the materials. The cost of an automated temperature monitoring system using IR cameras is
a modest and worthwhile investment.
System design can take the same form as
the one described earlier for hazardous waste barrels. Cameras are congured
to generate a direct alarm output to an
operator when user-dened maximum
temperature thresholds are exceeded.
Audible and visual alarms in a control room draw the operator’s attention to a possible spontaneous re development. Various types of software have been developed to isolate trouble spots, such as the waste pile zone monitoring system depicted in Figure 7.
Although self-ignition usually starts within the bottom layers of a stock pile,
continuous monitoring of the surface
reveals hot spots at an early stage (Figure
8), so measures can be taken to prevent a major re from breaking out. Large
storage yards generally require multiple
cameras for total coverage, with the cameras mounted on metal masts above
the stock piles. This calls for cameras with housings and other features designed
for reliable operation in harsh industrial
environments.
Critical Vessel Monitoring (CVM). There
are several applications where the temperature of a vessel and its contents are critical. The vessels could be used for chemical reactions, liquid heating, or merely storage. For large vessels,
the use of contact temperature sensors
poses problems. One reason could be non-uniform temperatures throughout a vessel and across its surface. This would
require a large number of contact type
sensors, whose installations can become
quite costly.
For most CVM applications, a few IR cameras can image nearly 100% of a vessels surface (Figure 9). Moreover, they
can measure the surface temperature of the CVM to trend and predict when the internal refractory will break down and compromise the mechanical integrity of
the system. If specic regions of interest (ROIs) must be focused on, IR camera rmware (or external PC software) allows
the selection of spot temperature points or areas for measurement.
Again, some variation of the systems
described earlier can be used. Depending
Figure 7. Control room for waste pile processing, and screen capture of the zone monitoring layout, which uses a FLIR IR camera on a pan-tilt mount for re hazard warning.
13
Remote IR Monitoring
Figure 8. Visible light and IR images of a coal pile – the thermographic image clearly identies a hot spot that is a re about to erupt.
on the application environment, an
explosion proof housing for the camera
may be a requirement. HMI (human­machine interface) software, such as SCADACAM iAlert from Pivotal Vision, can be used to provide a monitoring overview. This has the ability to combine
all of the camera images into a single spatial representation of the monitored
area – in this case, a attened-out view of the vessel. This view can be updated continuously for a near-real-time
thermographic representation.
Electrical Substation Monitoring. Reliable operation of substations is crucial for
uninterrupted electrical service. Besides lightning strikes and large overloads,
aging equipment and connections are a major cause of infrastructure failures
and service interruptions. Many of these failures can be avoided with eective preventative maintenance monitoring. Often, the temperatures of transformers, breakers, connections, etc. will begin to
creep up before a catastrophic failure occurs. Detection of these temperature increases with IR cameras allows
preventative maintenance operations
Figure 9. CVM monitoring example showing camera locations, network connections, and PC.
1 Computer 2 CAT-6 Ethernet cable with RJ45
connectors
3 Industrial Ethernet switch with PoE 4 ThermoVision™ A320 cameras 5 Industrial process to be monitored,
e.g., a gasier
14
Chapter 
before an unplanned outage happens.
(See Figure 10.)
The cameras can be installed on a pan/ tilt mounting mechanism to continually
survey large areas of a substation (Figure
11). A few cameras can provide real-time coverage of all the critical equipment
that should be monitored. In addition
to preventative maintenance functions, these cameras also serve as security
monitors for intrusion detection around the clock.
By combining the cameras’ Ethernet and/or wireless connectivity with a web-enabled operator interface, live
images can be transmitted to utility
control rooms miles away. In addition,
trending software can be used to detect dangerous temperature excursions and
notify maintenance personnel via email
and snapshot images of the aected equipment.
These features and functions are already in place at leading utility companies in
the U.S., such as Exel Energy’s “Substation
of the Future.” Companies such as
Exel consider IR monitoring a strategic
investment in automation, which is part of a common SCADA (Supervisory Control And Data Acquisition) platform
for maintenance and security operations.
The most advanced systems provide time-stamped 3-D thermal modeling of critical equipment and areas, plus temperature trending and analysis. A company-wide system of alerts provides alarms on high, low, dierential, and
ambient temperatures within or between
zones in real time.
The previous examples represent just a few applications that can benet from remote IR camera monitoring. A few
other applications where IR temperature monitoring is being used include:
Oil and gas industries (exploration • rigs, reneries, are gas ues, natural gas processing, pipelines, and storage facilities)
Electric utilities (power generation • plants, distribution lines, substations, and transformers)
Figure 10. Visible light and IR images of a substation showing a transformer with excessive temperature.
15
Remote IR Monitoring
Figure 11. Example pan/tilt mounting system.
Smarter surveillance for a smarter grid
Meet ScadaCam Intelligent Surveillance, the only system in its price range that can automatically perform site patrols, monitor equipment temperature, and scan for security breaches without human supervision.
By combining visual, thermal imaging, and thermographic cameras into a multifunctional operations and security automation tool, ScadaCam can detect, validate, and alarm you of problems that could otherwise result in a major outage – before they occur.
See it in action at www.pivotal-vision.com/tryit
16
Chapter 
Predictive and preventative • maintenance (continuous/xed
position monitoring of critical
equipment)
Besides these, there are many qualitative
remote monitoring applications where imaging is the predominant feature. For
example, IR cameras can be used as part
of an early warning system for forest
res (Figure 12), detecting blazes before signicant amounts of smoke appear. Another example is using IR imaging to look through condensation vapor that would otherwise obscure an operator’s view of equipment and processes. This is being used in coking plants, veneer mills, and plywood log handling operations, among others (see Chapter 1, Figure 1).
Summary
As noted in the text, IR camera
temperature data may be used for
qualitative monitoring or for quantitative
temperature measurement and control.
In the former, thermal images are
obtained and interpreted based on temperature contrast. It can be used to identify image areas that correlate
to sub-surface details, liquid levels, refractory, etc.
Quantitative measurements generally
require the IR camera to accurately determine the temperature dierence between the target object and its
surroundings. In remote monitoring, this
allows the temperature data to be used
for alarm purposes or to even shut down
equipment. Since temperature changes
slowly in many situations, the near-real-
time data communications of smart IR cameras are more than adequate for alarm and control systems.
Figure 12. Ngaro’s IRIS® Watchman forest re early warning system uses a FLIR IR camera.
17
Temperature Measurement
for Automated Processes
Temperature Measurement
for Automated Processes
Background
In Chapter  the emphasis was on
specic applications where a single
temperature threshold is programmed
into an IR camera, and when the
threshold is reached an alarm is triggered through a PLC. Multiple
cameras are often required, but viewing an IR cameras’ thermographic image is a secondary consideration – to verify an alarm condition. Chapter 3
focuses on applications where multiple
temperatures within a single camera’s FOV are important, and that information
is used for some sort of process control
function. In these applications, the
camera is typically integrated with other
process control elements, such as a PC or
PLC using third party software and more sophisticated communication schemes.
Typical Camera Measurement Functions
Many IR cameras provide the user with
dierent operating modes that support correct temperature measurements
under various application conditions.
Typical measurement functions include:
Spotmeter
Area
Image mask Delta T Isotherm Temperature range Color or gray scale settings
The last two are used with the others to
provide a visual indication of the range of temperatures in the camera’s FOV. Generally, spot and area temperatures
tend to be the most useful in monitoring
and control applications, and most
cameras allow multiple spots or areas to be set within the thermographic image.
For example, the FLIR A320 camera
supports up to four spots and four areas.
Cursor functions allow easy selection of
an area of interest, such as the crosshairs of the spot readings in Figure 1. In addition, the cursor may be able to select circlular, square, and irregularly shaped
polygon areas.
Figure 1. IR image of a printed circuit board indicating three spot temperature readings. Image colors correspond to the temperature scale on the right.
The spotmeter nds the temperature
at a particular point. The area function isolates a selected area of an object or
scene and may provide the maximum, minimum, and average temperatures
inside that area. The temperature measurement range typically is
selectable by the user. This is a valuable
feature when a scene has a temperature
range narrower than a camera’s full-
scale range. Setting a narrower range allows better resolution of the images and higher accuracy in the measured
Chapter 3
18
Chapter 3
Figure 2. Gray scale images of a car engine – the left view has white as the hottest temperature and the right view shows black as the hottest.
temperatures. Therefore, images will
better illustrate smaller temperature
dierences. On the other hand, a
broader scale and/or higher maximum temperature range may be needed to
prevent saturation of the portion of the
image at the highest temperature.
As an adjunct to the temperature range selection, most cameras allow a user
to set up a color scale or gray scale to
optimize the camera image. Figure 2
illustrates two gray scale possibilities.
In Figure 1, a so-called “iron scale” was
used for a color rendering. In a manner
similar to the gray scale above, the
hottest temperatures can be rendered as either lighter colors or darker colors.
Another possibility is rendering images
with what is known as a rainbow scale
(Figure 3).
While choice of color scale is often a
matter of personal preference, there may
be times when one type of scale is better than another for illustrating the range of temperatures in a scene.
Figure 3. Rainbow scale showing lower temperatures towards the blue end of the spectrum.
Application Examples
Go/No-Go. In these applications, one or
more temperatures are monitored to
make sure they meet process criteria,
and machinery is shut down or product rejected when a measured temperature
goes above or below the setpoint. A
good example of this is a manufacturer
of automotive door panels that uses
IR cameras to monitor and measure part temperatures prior to a molding procedure.
19
Temperature Measurement for Automated Processes
This process starts with reinforcing parts
that have been stored in a warehouse.
In either the warehouse or during
transport to the molding line, these
parts can become wet due to moisture condensation or exposure to inclement
weather. If that happens, they may not
reach a high enough temperature in the
molding press and nished panels will be
of poor quality.
The parts go into the press two at a
time from a conveyor where they are sealed together and the nished door
panel is molded into the required shape
for a specic car model. If the parts are wet, this creates steam in the press and
causes mold temperature to be too low.
However, it was found that movement of wet parts on the conveyor causes their temperature to be lower than normal. So, just before the parts go into the press, the conveyor stops and an IR camera makes a non-contact measurement
of their temperature. The diagram in
Figure 4 is typical for this type of quality
control application.
The IR camera’s area tools are applied to
the thermographic image to check for the minimum allowable temperature of the two parts. If either temperature is
below the setpoint (typically, the ambient temperature), then a digital I/O output to
a PLC causes an alarm to be sounded and
Figure 4. Typical Go/No-Go QC inspection system using IR cameras.
1 Computer or PLC 2 CAT-6 Ethernet cable with
RJ45 connectors
3 Industrial Ethernet switches
with ber optic ports 4 Fiber optic cable 5 ThermoVision™ A320 or A325
cameras
6 Industrial process to be
monitored, e.g., items on a
conveyor belt
20
Chapter 3
the molding line is halted so the parts can
be removed.
For OEMs, preventing bad panels from getting to the end product avoids a
potential loss of business. Warranty replacement of a door panel after an end customer takes possession of the car is an
expensive proposition for the OEM.
The trick is to make sure the camera is measuring the temperature of the parts
and not the oor beneath the conveyor, which is within the camera’s FOV and
typically much cooler. This occurs when
the parts are not in the proper position. A
photoelectric detector tells the PLC when the parts enter the press area; otherwise its ladder logic ignores the alarm output from the camera.
Continuous Process Monitoring.
Temperature is an important variable
in many processes. It can either be an integral part of a process or act as a proxy for something else. The following describes an example that encompasses both of these situations.
Articial ber production typically involves a continuous extrusion process.
Multiple strands may be extruded
simultaneously or, in the case of non­woven sheets, a web process may be involved. In either case, monitoring
the temperature of the material as it comes out of the extruder can detect strand breakage or material blockage
and backup in the process. Using an IR
camera for unattended monitoring can
catch these malfunctions early, before
a huge mess is created that causes a long machinery outage and costly
production losses. In addition, the actual
temperature readings can be used for trend analysis.
Depending on the application, either the
spot or area measurement functions of
the camera can be used. In the latter case,
it is likely that the application would take
advantage of all the area measurement capabilities – minimum, maximum, and average temperatures of the dened
area. If any of these were to fall outside
the user-dened limits, the application
program running on a PC or PLC could instantly shut down the process machinery.
In one such application, FLIR customized the camera rmware to
allow simultaneous monitoring of up
to 10 dierent areas. Figure 5 shows a monitored area covering six ber strands coming out of the extruder, along with an
alarm setpoint temperature in the upper left corner.
Figure 5. Monitoring of articial bers coming out of an extruder.
As in the case of many remote monitoring applications, the user may choose to route the camera’s analog video to a
control room monitor. For cameras
with an Ethernet connection, digitally
21
Temperature Measurement for Automated Processes
compressed (MPEG-4) streaming video can be available for monitoring on a PC screen. With FLIR’s A320 camera, images and alarms can be sent to a remote PC via TCP/IP and SMTP (email) protocols.
While a visible light camera may be able to detect broken ber strands, an IR camera can also provide temperature
measurements for trending and statistical
process control (SPC) purposes. In addition, some textile processes create steam or condensation vapors that a visible light camera cannot see through, but an IR camera can. Thus, an IR camera provides multiple functions and is more cost eective.
Data Communications and Software Considerations
Dierent cameras have dierent video frame rates. The frame rate governs how
frequently the thermographic image and its temperature data are updated.
A typical rate might be every 200ms or so. The camera’s digital communications
protocol could create a small amount of additional latency in the update process.
Still, because process temperatures tend to change slowly, collecting temperature data at this rate provides a wealth of
information for quality control purposes.
In many IR cameras there is some sort of serial/socket interface that can be used for communications with the PC
or PLC that is running a control script,
or application. When a system designer
or user is most familiar with PLCs, the
control algorithm can be built around
a virtual PLC created on a PC, which
emulates actual PLC hardware and logic.
In any case, a human-machine interface (HMI) is created to monitor data coming
from the camera. The details described
below are based on FLIR’s A320 camera, but should be representative of most cameras that transmit data over an
Ethernet link.
The only physical interface for digital
data transfer from the FLIR A320 is the Ethernet port. Only TCP/IP is supported,
but the camera should work seamlessly
on any LAN when the proper IP address, netmask, and possibly a gateway is set
up in the camera. The two main ways of controlling the camera are through the command control interface and the resource control interface. Digital image
streaming, data le transfer, and other functionality is provided through the IP services interface. A lot of software
functionality is exposed through software resources. These resources can be reached through the FLIR IP
Resource Socket Service. This is the camera’s resource control (serial/socket)
interface. Independent of the physical
Ethernet interface, it is possible to access
the camera system using TCP/IP with
telnet, ftp, http, and FLIR Resource Socket Services (among others).
Most PLCs provide serial/socket interfaces for Ethernet. One example is Allen­Bradley’s EtherNet/IP Web Server Module (EWEB for short). Another example is HMS Industrial Network’s Anybus X-Gateway Ethernet interface module, which can convert this serial socket interface to many industrial network protocols, such as EtherNet I/P, Modbus-TCP, Pronet, Ethernet Powerlink, EtherCAT, FLNet., etc.
Camera setup and data acquisition is normally done directly through the FLIR IR
MONITOR and IP CONFIG software running on a PC. Afterward, the camera can be
22
Chapter 3
connected on the network for continuous
monitoring and data logging via PC or PLC control. Typically, the telnet protocol,
accessed by the Windows® PC running
the application program, is used to query
the camera for data. This protocol is also
available for most PLCs. In either case, this takes place though the camera’s Resource Socket Services. (Command syntax is contained in the camera’s ICD manual; a few examples are listed in Appendix D.)
The system designer or FLIR would create the message instructions that allow the PLC to query the camera for temperature data and thermographic
images in the same way it is done with PC
control. Alternatively, the PLC can hold
the Ethernet port open and call for the camera to continuously output data to this port at the maximum rate possible. In
either case, alarm functions and decision-
making is performed by the application
program running on the PLC (or PC if applicable). (See Figure 6.) Typically,
temperatures and images collected for trend analysis and statistical process control purposes are stored on a separate
server connected to the network, which
is running transaction manager software for downloading and storing data.
1 Computer, PLC, and/
or transaction manager
server
2 CAT-6 Ethernet cable
with RJ45 connectors
3 Industrial Ethernet
switches with ber optic
ports
4 Fiber optic cable
 Wireless access points
6 CAT-6 Ethernet cable
with RJ45 connectors.
Powering the camera
using PoE (Power over Ethernet)
7 Industrial Ethernet
switch
8 ThermoVision A320
cameras monitoring a process or other target objects
Figure 6. Generalized IR machine vision system and its communications network
23
Temperature Measurement for Automated Processes
Figure 7. Example of a control and data acquisition option for IR cameras
©2008 National Instruments Corporation. All rights reserved. Nati onal Instruments, NI, and ni.com are trademarks of National Instruments. Other product and company names listed are trademarks or trade names of their respective companies. 2008-9522-221-101
>>
Get your free 30-day trial version at ni.com/vision/vbai
800 891 8841
Inspect Faster
with NI Vision Builder AI
National Instruments Vision Builder for Automated Inspection (AI) now features an innovative state machine editor, taking you from initial design to deployed vision application faster than ever. Easily integrate with existing industrial control hardware no programming required.
NEW! Vision Builder AI
Configurable Machine Vision
No Programming Required
Acquisition from Thousands of Cameras
2008-9522-221-101-D 4/29/08 10:25 AM Page 1
24
Chapter 3
For system developers who are writing or modifying code with Visual Basic, C++, etc. for customized applications running on a PC, there are a few options. FLIR’s Researcher package supports OLE-2,
the Microsoft standard for linking and embedding data between applications. Image and temperature data can be linked from Researcher into other
compliant applications, such as Excel. The linked data updates automatically, so if a temperature value changes in Researcher
it will automatically change in the linked
application. In addition, Researcher provides an automation interface that
can be used to control the software using
Visual Basic or VBA. Other o-the-shelf options for OLE control include National Instruments’ MATLAB and LabVIEW®. However, none of the aforementioned are OPC (OLE for Process Control) compatible.
There are other out-of-the-box solutions
that do not require the writing of
application source code. One of these is IRControl from Automation Technology, GmbH. IRControl simplies automated
processing of complex tasks with its
built in Automation Interface based on
Microsoft® COM/DCOM. All essential
measurement, analysis, and control
functions for FLIR IR cameras are directly programmable using macro commands. This allows the execution of control scripts automatically based on digital
input events. In addition, IRControl
accepts remote control commands sent
over an RS-232 link. Therefore, remote
control of IRControl by other computers
or PLCs is greatly simplied. The software also includes a comprehensive report
generator.
Summary
A variety of control and data acquisition options are available for IR cameras (see Figure 7). They are similar to those used with visible light cameras that are employed in machine vision and automation systems. IR cameras provide the added advantage of accurate non-
contact temperature measurements within a single instrument.
25
Combining Machine Vision and
Temperature Measurement
Combining Machine Vision and Temperature Measurement
Background
Traditionally, visible light cameras have been a mainstay in machine vision
systems used for automated inspection and process control. Many of these systems also require temperature measurements to assure product quality.
In numerous cases, an IR camera can
supply both an image of the product and critical temperature data. If the
application will not benet from thermographic images and non-contact temperature measurements, then a visible light camera is certainly less expensive. If the opposite is true, then an
IR camera should be considered by the system designer.
As the sophistication of IR cameras continues to increase, along with associated hardware and software, their
use in automated systems is growing rapidly. Because of their combined imaging and temperature measurement
capabilities, they can be very cost eective. The main impediment to their wider usage is system designers’ lack
of familiarity with IR camera features
and the related standards, systems, and
software that support them. This chapter supplies a good deal of that information.
Machine Vision Applications
As in the case of visible light cameras,
thermographic cameras and their
associated software can recognize the size, shape, and relative location of target objects (i.e., they can do pattern
matching). Moreover, the electronics in newer IR cameras provide fast signal processing that allows high video frames rates (60Hz or higher) to capture relatively fast-moving parts on a production line. Their A/D converters combine short integration times with 14- to 16-bit resolution, which is critical for properly characterizing moving targets or targets
whose temperatures change rapidly.
Figure 1. Results of automated inspection of ICs on a circuit board
One example of the latter is automated
inspection of operating ICs on a circuit
board (Figure 1). In some cases, this involves overload testing in which an
IC is subjected to a current pulse so its
heat loading can be characterized. In one such case the IC is forward and reverse biased with current levels outside of
design limits using a pulse that lasts
800ms. The IR camera captures images
during and after the current pulse to
characterize temperature rise and fall. With a 60Hz frame rate, a new frame can be captured about every 17ms. In such a system nearly 50 frames can be captured during the 800ms pulse, and many more
Chapter 4
26
Chapter 4
are typically captured afterward to reveal
heat dissipation characteristics.
In other applications of this sort, a good
image can be stored and compared to
the inspection image by using pixel-by­pixel subtraction. Ideally, the resulting image would be entirely black, indicating no dierence and a good part. Areas with excessive temperature dierences indicate a bad part, making it very easy to
discern unwanted dierences.
There are many other applications
where the combination of non-contact
temperature measurements and imaging
at high frame rates is extremely valuable.
Some automated systems where IR cameras are already being used include:
Automotive part production and
assembly lines
Steel mill operations, such as slag
monitoring and ladle inspection
Casting, soldering, and welding of
metals and plastics Food processing lines Product packaging
Non-destructive testing, like sub-• surface detection of voids in molded
parts Electric utility equipment monitoring
R&D, prototyping, and production in
the electronics industry
An interesting automotive example is
monitoring the temperature distribution
of a pressure casting mold for a safety­critical part (Figure 2). Prior to installation of the IR machine vision system, the manufacturer was doing 100% inspection using an X-ray system to reveal
subsurface imperfections. It was not
practical to do this as an inline procedure,
so the X-rays were taken a few hours after part production. If the X-rays showed a signicant problem in parts coming from a particular mold, this information was
relayed to the production area so that mold temperatures could be adjusted. This was a lengthy and costly process that often resulted in high scrap rates. With
the IR camera system, the mold operator
can immediately check and adjust the temperature distribution of the mold.
Figure 2. Pressure casting mold and its temperature distribution – an IR camera image is used by the operator to adjust the mold temperatures as required to produce good parts.
Enabling Technology
Data communications are the backbone
of modern industrial SCADA, PLC, HMI’s,
and Machine Vision systems. Ethernet has become the de facto standard for such
systems. Considering this, the features of
27
Combining Machine Vision and Temperature Measurement
IR cameras that make for practical use in
machine vision applications are Gigabit Ethernet (GigE) connectivity, GigE Vision™ compliance, a GenICam™ interface, and
a wide range of third party software that supports these cameras. There are other hardware features that are also important.
Generally, ultra-high detector resolutions
are not needed in the targeted
applications, so a typical focal plane array (FPA) would be 320x240 pixels. Nevertheless, outputting a 16-bit image stream of these 76,800 pixels at a 60Hz frame rate amounts to about 74Mb/
sec. While this is much slower than a
1000baseT Ethernet system is capable of,
multiple cameras may be connected and
there may be a lot of other trac on the
network between image transmissions.
To speed up image transfers, data analysis and decision-making must take
place outside the camera and is one of the reasons why there is a good market
for third-party thermographic software.
The other reason is that most machine
vision systems are custom designed for specic production processes. Of course, IR camera manufacturers supply various types of software to support their
products and facilitate application in these systems.
The goal of the GigE Vision technical standard is to provide a version of GigE
that meets the requirements of the
machine vision industry. One of the industry objectives is the ability to mix and match components from various
manufacturers that meet the standard.
Another is relatively inexpensive accessories, such as cabling, switches, and network interface cards (NICs) as well
as the ability to use relatively long cable
runs where required.
The GigE Vision standard, which is based on UDP/IP, has four main elements:
A mechanism that allows the
application to detect and enumerate
devices and denes how the devices obtain a valid IP address.
GigE Vision Control Protocol (GVCP) • that allows the conguration of detected devices and guarantees
transmission reliability.
GigE Vision Streaming Protocol (GVSP) • that allows applications to receive information from devices.
Bootstrap registers that describe the • device itself (current IP address, serial number, manufacturer, etc.).
With GigE capabilities and appropriate software, an IR machine vision system
does not require a separate frame
grabber, which was typically the case with visible light cameras in the past. In eect, the GigE port on the PC is the frame grabber. Older visible light cameras that have only analog video outputs (NTSC and PAL) are limited to much lower frame rates and video monitor observations. By using GigE, an IR vision system not only has higher frame rates, but can be monitored remotely over
much greater distances compared to local processing and transmitting data
over USB, Firewire, CameraLink, etc. In addition, Ethernet components are inexpensive compared to frame-grabber
cards and related hardware.
A GigE Vision camera typically uses an NIC, and multiple cameras can be connected on the network. However, the drivers supplied by NIC manufacturers
28
Chapter 4
use the Windows or Linux IP stack, which may lead to unpredictable behavior,
such as data transmission delays. By
using more ecient dedicated drivers compatible with the GigE Vision standard, the IP stack can be bypassed
and data streamed directly to memory at
the kernel level of the PC system. In other words, Direct Memory Access (DMA) transfers are negotiated, which also eliminates most CPU intervention. Thus a near-real-time IR vision system is created in which almost all of the CPU time is
dedicated to processing images.
To make sure a camera is GigE Vision compliant, look for the ocial stamp (shown in Figure 3) that can only be
applied if the camera conforms to the standard.
Figure 3. Ocial trademark for GigE compliant products
GenICam compliance should also
be considered for an IR camera.
GenICam compliance makes it easier for developers to integrate cameras into their IR vision system. The goal of the GenICam standard is to provide
a generic programming interface for
all kinds of cameras. No matter what interface technology (GigE Vision, Camera Link, 1394, etc.) is used, or what camera features are being implemented,
the application programming interface
(API) should be the same. The GenICam
standard consists of multiple modules and the main tasks each performs are:
GenApi: conguring the camera• Standard Feature Names:
recommended names and types for common features
GenTL: transport layer interface,
grabbing images
The GenApi and Standard Feature Names modules are currently part of the standard module only. GenTL should be nished soon.
Common tasks associated with IR
cameras in machine vision systems include conguration settings, command and control, processing the image, and
appending temperature measurement results to the image data stream. In
addition, the camera’s digital I/O can be used to control other hardware, and there are triggering and synchronization functions associated with real-time data acquisition. GigE Vision makes hardware independence possible, while GenICam
creates software independence. For
example, in a system with IR cameras
compliant in both and connected to a
GigE network, virtually any application
program can command a camera to
send a 60Hz stream of images that can
be easily captured without dropping frames and losing important data. This information can be processed for alarm
functions, trend analysis and statistical
process control.
Third Party Software Expands
Applications
By adhering to the standards described
above, IR camera manufacturers are making it easier for developers to integrate their cameras into vision
systems with a broad array of functions
29
Combining Machine Vision and Temperature Measurement
#USTOM)MAGING3OLUTIONS
4URNKEY3OLUTIONSFOR
s0ROCESS#ONTROL s0ROCESS-ONITORING s.ON$ESTRUCTIVE4ESTING s#USTOM3YSTEMS
4URNKEY3OLUTIONSFOR
s0ROCESS#ONTROL s0ROCESS-ONITORING s.ON$ESTRUCTIVE4ESTING s#USTOM3YSTEMS
#RITICAL6ESSEL-ONITORING
0ACKAGE3EALING7ASTE"UNKER-ONITORING0ETROCHEMICAL0ERIMETER3ECURITY0LANT-ONITORING3OLAR0ANEL)NSPECTION#OMPOSITE-ATERIAL4ESTING
#RITICAL6ESSEL-ONITORING0ACKAGE3EALING7ASTE"UNKER-ONITORING0ETROCHEMICAL0ERIMETER3ECURITY0LANT-ONITORING3OLAR0ANEL)NSPECTION#OMPOSITE-ATERIAL4ESTING
WWWMOVIMEDCOM
s s s s s s s
s s s s s s s
(Figure 4). Camera manufacturers also supply a variety of software products to ease integration tasks. For example, the FLIR A325 comes with three packages that
run on a PC controller:
IP Conguration utility – nds cameras • on the network and congures them
IR Monitor – displays images and
temperature data on up to nine cameras simultaneously
AXXX Control and Image interface • – low-level descriptions of how to communicate with the camera, including image formats and C-code
examples
In addition, optional software developer toolkits are available (FLIR SDK, LabVIEW SDK, Active GigE SDK from A&B Software, etc.) for those creating source code for
custom applications within programming
environments such as Visual Basic, C++, Delphi, etc. However, the strength of a camera like the A325 is its ability to
interface with third party software that
eliminates or minimizes the need to write source code. For example, National Instrument’s Vision Builder for Automated Inspection is a congurable package for building, benchmarking, and deploying machine vision applications (Figure
5). It does not require the user to write program code. A built-in deployment
interface facilitates system installation
Figure 4. IR cameras can be used in a broad array of applications
30
Chapter 4
and includes the ability to dene complex pass/fail decisions, control digital I/O, and communicate with serial or Ethernet devices, such as PLCs, PCs, and HMIs. Similar features are available in Common Vision Blox, a Stemmer Imaging product that contains hardware- and language-independent tools and libraries
for imaging professionals.
By using third party software to get much
of the analytics, command, and control
functions out of the camera and onto a
PC, application possibilities are greatly expanded. One possibility is creating a mixed camera system. For instance, IR
cameras could be used to supply thermal
images and temperature data, while visible light cameras could provide “white
light” color recognition.
The food processing industry is one in
which higher level analytics are used with IR cameras for automated machine vision applications. A broad area of applications where IR vision systems excel is in 100%
inspection of cooked food items coming
out of a continuous conveyor oven. A
primary concern is making sure the
items have been thoroughly cooked, which can be determined by having the camera measure their temperature, which is illustrated in Figure 6 for
hamburger patties. This can be done by
dening measurement spots or areas
corresponding to the locations of burgers
as they exit the oven. If the temperature of a burger is too low, the machine vision program logic not only provides an alarm, but also displays an image to the oven operator to show the specic burger that should be removed from the line. As in other applications, minimum, maximum, and average temperatures can be collected for specic burgers or the FOV as a whole and used for trending and
SPC purposes.
Figure 6. IR machine vision image for checking hamburger doneness by measuring temperature
In another example involving chicken tenders, temperature is again used to
check for proper cooking. The pieces
come out of the oven and drop onto another conveyor in more or less random locations (Figure 7). The operator can
use the thermographic image to locate undercooked items within the randomly
spaced parts and then remove them from the conveyor.
In the production of frozen entrées, IR machine vision can use pattern
recognition software to check for proper
lling of food tray compartments.
Find any number of edges
Set up coordinate systems
Find and match patterns
Acquire with IEEE 1394 and GigE cameras
Detect and measure objects
Calibrate measurements to real-world units
Perform advanced geometric analysis
Find and measure straight edges
Find circular edges
Make caliper distance measurements
Make pass/fail decisions
Read text (OCR)
Communicate with external devices suc
h
as PLCs
Measureintensity
Read 1D and 2D bar codes
Figure 3. Examples of the many functions available in Vision Builder for automated inspection
31
Combining Machine Vision and Temperature Measurement
Similarly, it can be used for 100% inspection of the heat-sealed cellophane cover over the nished entrée. An added
function could be laser marking of a
bad item so it can be removed at the
inspection station.
Summary
IR machine vision and temperature
measurements can be applied to an
innite number of automated processes. In many cases, they provide images and information that are not available with visible light cameras, and they
also complement white light images where the latter are required. IR cameras
like the FLIR A325 provide a stream of digitized IR images at fast frame rates for relatively high-speed processes, which can be transmitted over GigE networks to remote locations. Compliance with GigE Vision and GenICam standards means
that such cameras can be integrated
with a wide variety of similarly compliant
equipment and supported by a broad range of third party software. Trigger and
synchronization capabilities allow them to control, or be controlled by, a host of other types of equipment. The availability of wireless and beroptic line adapters
allow these cameras to be used almost
anywhere, including over long distances.
Figure 7. An IR temperature measurement and thermographic image are used to locate undercooked chicken tenders and stop the line so bad parts can be removed.
32
Chapter 
Real-Time Control Issues
Background
Real-time control is an important issue in most IR machine vision systems used for
automated temperature monitoring and
inspection. Having said that, it should be noted that real time tends to be a relative term, the measure of which varies with
the application and user requirements. In
some applications, users would consider a response time of 100 milliseconds to meet their denition of real-time. On the other hand, many electronic events are extremely fast or short-lived, and a one-microsecond response might be needed. As mentioned in earlier chapters,
process temperatures tend to change
relatively slowly, so an IR machine vision
system that can update images and
temperatures every 10-100ms, or even less frequently, may be adequate.
Hardware and Software Platform Considerations
In most cases, a PC with a Microsoft Windows operating system (OS) isn’t well suited for controlling fast, real-time
applications. Windows is referred to as a
non-deterministic OS because it typically cannot provide predictable response
times in critical measurement and control
situations. Therefore, the solution is to
link the PC to a system that can operate
autonomously and provide rapid,
predictable responses to external stimuli.
Deterministic applications (those intended to be event driven) are
controlled better with systems based on an embedded microprocessor and/
or digital signal processor (DSP) that has a dierent type of OS – or perhaps a special version of Windows other than
the ones typically found on a home or
Figure 1. PLCs are a good choice for creating deterministic (event-driven) systems, supported by a PC that is used for data trending.
GPIO or PLC
33
Real-Time Control Issues
oce PC. Often, a system based on a PLC with 115VAC control I/O is much more appropriate for real-time applications (Figure 1). PLC processors are designed
to operate with deterministic control
loops, and the 115VAC control signals are
inherently immune to noisy industrial
environments.
If the required response time is long
enough, and there are other reasons
to use one of the familiar Windows
operating systems, keep in mind that special steps are needed to improve its
data polling methodology. In a polled
system, the PC checks many devices to see if they’re ready to send or receive
data. In the context of data acquisition
from an Ethernet-based IR camera, this typically involves reading values from a data stream. In a Windows-based PC, the time between polled readings is scheduled by Windows, so it’s non­deterministic. In other words, the time at
which Windows will initiate an operation cannot be known precisely. Its operation depends on any number of system
factors, such as computer speed, the OS, programming languages, and application code optimization.
Polling can be appropriate with slower, less time-sensitive operations. In contrast, event-driven programming schemes are less dependent on OS timing and tend
to reduce latency problems. They can be used to create more deterministic
systems that collect discrete data values
that are closely related to the physical phenomena being represented.
Creating such a system within a
Windows OS environment generally
requires writing program code using
Visual C/C++ , Visual Basic, etc. Using
these tools, a programmer can take advantage of Windows events and
messaging functionality to create a more deterministic application that
runs relatively fast and provides tight
control. Rather than constantly polling to
determine if data is ready for collection, such programs can use the PC’s CPU for additional tasks, such as database or network access, until interrupted
by the automation system hardware.
As discussed in Chapter 4, there are software developer kits that take some of the work out of these tasks, and third
party software packages can eliminate
or minimize the need to write program code. An example is illustrated in Figure 2.
Figure 2. Third party software provides powerful control and analytic tools for IR machine vision systems without writing program code.
Data Communication Latencies
Hardware and data communications
have signicant eects on system
response time. The Ethernet interfaces on many IR cameras allow communication
distances of 4000 feet or more. Wireless and beroptic adapters and hubs can extend the scope of the network even more. Networked systems require the
34
Chapter 
installation of one or more NICs in the PC and conguring its OS for network
support. These requirements are easily
and economically met with Ethernet, TCP/IP, and Windows, as described in Chapter 3.
A functional drawback of Ethernet-based systems concerns real-time control. Like Windows, Ethernet is a non-deterministic
system that in many applications
precludes fast, real-time process control. This can become even more of
an issue when the World Wide Web is
involved. Again, there are work-arounds to minimize inherent weaknesses. As
mentioned in Chapter 4, drivers supplied by NIC manufacturers use the Windows or Linux IP stack, which may result in data
transmission delays. By using dedicated
drivers compatible with the GigE Vision standard, data can be streamed directly to memory using a DMA transfer.
Since older communication protocols
(RS-232, 422, 485, etc.) are even slower,
Ethernet is still the protocol of choice
in most IR machine vision systems. The digitized streaming video from FLIR’s A325 camera allows near-real-time
data acquisition of thermal images
and temperature data – provided the
1 Computer, PLC, and/
or transaction manager
server
2 CAT-6 Ethernet cable
with RJ45 connectors
3 Industrial Ethernet
switches with ber optic
ports
4 Fiber optic cable
 Wireless access points
6 Industrial Ethernet
switch
7 ThermoVision A320 or
A325 cameras monitoring
a process or other target objects
Figure 3. Generalized IR machine vision system and its communications network
35
Real-Time Control Issues
PC has the appropriate NIC driver and application program. Network adapters for beroptic and wireless connectivity can extend Ethernet’s scope (Figure 3). Other hardware timing issues can be minimized by using direct-wired digital I/O and triggering between individual cameras, PLCs, etc. Analog video (NTSC and PAL) for conventional image
monitoring is probably most applicable
to qualitative applications where timing
is not critical.
IR Camera Hardware and Firmware Issues
Thermal Time Constants for Cooled and Uncooled IR Cameras. In general, time
constant refers to the time it takes for a sensing element to respond to within
63.2% of a step change in the state of a target that is being sensed (Figure
4). In IR sensing and thermography,
the thermal time constant of an IR
camera’s detector is a limiting factor in
instrument performance as it relates to response time.
100%
80%
63%
0 1 2
Thermal Time Constants
Step Change in Target State
Figure 4. Thermal time constant concept showing an integral number of time constants on the X-axis.
Older IR cameras have response times similar to the human eye, so they are
unsuitable for capturing thermal images
of fast moving objects or those with rapidly changing temperatures. Newer IR cameras have detectors and digital
electronics with response times in the
sub-millisecond region. Cooled quantum detectors are very sensitive and very fast (sub-microsecond response times),
but their bulkiness and cost tends to rule them out of many automation
applications. In addition, quantum detectors have response curves with detectivity that varies strongly with IR wavelength. FLIR has made recent improvements to its uncooled
broadband microbolometer detectors
and associated A/D converters so they
can continuously output images with embedded temperature data at a
60Hz rate. This is satisfactory for most
temperature monitoring and IR machine
vision applications.
Temperature Measurement Range.
The overall temperature range of an
IR camera is primarily a function of its detector and calibration. Camera
electronics, which include calibration functions, can handle wide variations in absolute detector sensitivities. For example, the FLIR A325’s overall measurement range is divided into user­selectable temperature scales that have a measurement accuracy of ±2°C (±3.6°F) or ±2% of reading:
–20°C to + 120°C (–4°F to +248°F)• 0°C to +350°C (32°F to +662°F)• Optionally, 250°C to +1200°C (482°F to
2192°F)
This is a valuable feature when a scene
has a temperature range narrower than
36
Chapter 
a camera’s overall range. Selecting a
narrower scale allows better resolution of the images and higher accuracy in
the measured temperatures. Therefore,
the images will better illustrate smaller
temperature dierences. On the other hand, a broader scale and/or higher
maximum temperature range may be
needed to prevent saturation of the
portion of the image at the highest temperature.
It’s important to understand how the camera’s calibration and temperature
measurement processes aect its response time. IR cameras measure
irradiance, not temperature, but the
two are related. When an IR camera
is thermographically calibrated, it
can measure temperatures based on
standard blackbody radiances at specic temperatures. As will be discussed later, the emissivity of the target object being measured is vital to achieving accurate temperature readings. (Emissivity or emittance is the radiative properties of an object relative to a perfect blackbody.)
When an IR camera is calibrated at the
factory, calibration factors are stored internally as a table of values based on the camera’s A/D counts from the
temperature/radiance measurements of a standard blackbody. When the system
makes a measurement in an application, it takes the digital value of the signal at a given moment, goes into the appropriate calibration table, and calculates temperature. Before the nal result is presented, due consideration is given to other factors, like emissivity of the target objects, atmospheric attenuation, reected ambient temperature, and the camera’s ambient temperature drift.
Figure 5. Creating a temperature scale narrower than the camera’s full range improves image resolution and may improve defect detection.
As an adjunct to major temperature scale selections, most IR cameras allow a user
to set up a color scale or gray scale for a
temperature range that’s even narrower (Figure 5). This should be done where practical, not only because of improved image resolution, but also because of response time considerations. A narrower temperature range can reduce the A/D converter’s processing load and overall
response time of the system.
Another complexity is the fact that each individual pixel in the camera’s
focal plane array has a slightly dierent
gain and zero oset. To create a useful thermographic image, the dierent
gains and osets must be corrected
to a normalized value. This multi-step
calibration process is performed by the
camera rmware (Figure 6). The non­uniformity correction (NUC) factors are
also stored in a table.
IR cameras also have dierent
measurement modes: spotmeter and area measurements in the case of the
FLIR A320 Series. The spotmeter nds the
temperature at a particular point whereas the area function isolates a selected area
of an object or scene. In the latter case,
37
Real-Time Control Issues
camera rmware nds the minimum and
maximum temperatures and calculates
the average temperature inside the area selected. Clearly, more processing time is required for area measurements,
particularly if multiple areas are selected. This also means that more data is being
transmitted over a machine vision system’s communications network, along
with more latency.
Emissivity Calibration. Earlier, it was
pointed out that accurate temperature
measurements on a specic object
require the emissivity value for that object. In eect, this adjusts the factory
calibration that is based on a perfect
blackbody having an emissivity value of
1.0. This adjustment consumes processor time. To avoid this, the FLIR A325 uses a global emissivity value (input by the user) for the camera’s entire FOV. Normally, this isn’t a problem for machine vision applications and it avoids the time required to apply non-global emissivity values on the y. Instead, the application
program is set up to make decisions
Figure 6b. Final steps in IR camera’s NUC process
Signal SignalWithout any correction
First correction step
Radiation
–20°C+12 0°C
+20°C
Radiation
20°C +120°C
+20°C
Signal Signal
Third correction, Non-Uniformity Correction (NUC)
After NUC
Radiation
–20°C+12 0°C
+20°C
Radiation
20°C +120°C
+20°C
Figure 6a. First step in detector non-uniformity correction (NUC) performed by IR camera rmware
38
Chapter 
based on the temperature value of a target area compared to a standard value or compared to the target’s surroundings.
While this may not be accurate in an
absolute sense, it is the relative dierence
that is most important.
If the system developer wants accurate
temperature measurements on dierent
objects with dierent emissivities (e.g., for PC board inspections), then he/she must create an emissivity map for the camera’s FOV. This cannot be done with
the data coming out of a camera using
a global emissivity value. To create an emissivity map, the developer will need to write some program code, typically by using the FLIR Software Developers Kit. The routine that’s developed sets up
the system to read the FLIR proprietary
stream of data coming out of the A/D converter and applies emissivity values to it. This creates an emissivity map that covers selected areas or spots within the FOV.
Important Thermographic Principles
As alluded to above, there must be
a temperature dierence between a target object and its surroundings in order to create a useful thermographic
image. In most situations, the user also needs a measurement of this relative
temperature dierence for decision
making, either automatically or by a machine operator. However, there are a
few ambient conditions that may obscure the temperature dierence.
In addition to emitting radiation, an
object reacts to incident radiation from its surroundings by absorbing and reecting a portion of it or by allowing
some of it to pass through (as through a
lens). Therefore, the maximum radiation
that impinges on an IR camera lens aimed at an object comes from three sources:
(1) the object’s inherent temperature without inuence from its surroundings, (2) radiation from its surroundings that was reected onto the object’s surface, and (3) radiation that passed through the
object from behind. This is known as the
Total Radiation Law (see below). However,
all these radiation components become attenuated as they pass through the atmosphere on their way to the camera lens. Since the atmosphere absorbs
part of the radiation, it also radiates
some itself.
Total Radiation Law
1 = α + ρ + τ.
The coecients α, ρ, and
τ describe an object’s
incident energy absorption
(α), reection (ρ), and
transmission (τ).
Figure 7. Total Radiation Law
The role of emissivity in distinguishing an object’s temperature from its surroundings was discussed above. As alluded to earlier, it’s wise to take precautions that prevent reected energy
from impinging on the target. Corrections for atmospheric attenuation are normally
built into the camera rmware. Still,
other gases and hot steam between the target and the camera lens can make
measurements impossible, or at least inaccurate. Similarly, an object that is transparent to IR wavelengths may result
in the camera measuring the background behind it or some combination of the object and the background.
39
Real-Time Control Issues
In the last two cases, spectral lters that are selective at specic wavelengths can help. Certain lters can make an
otherwise opaque gas appear transparent or a transparent object appear opaque
over the appropriate IR band (Figure 8).
Summary
IR cameras used in machine vision and
other automation systems are analogous
to visible light cameras in similar systems. White light cameras have optical issues that must be managed, whereas IR cameras have thermographic issues to resolve. In both cases, achieving real-time
(or near-real-time) response requires
thoughtful selection of the controller and careful design of the application program. Third party software can
provide out-of-the-box program development tools that eliminate or minimize the need to write program code. Generally, there are no perfect solutions – developing an automated machine vision system, whether based on visible light or IR, usually involves
compromises of one sort or another. The camera manufacturer can be a
great source of help in developing
these systems.
Filter Adaptation
1.0
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0.0 3 3.1 3.2 3.3 3.4 3.5 3.6 3.7 3.8 3.9 4.0
Wavelength, µm
Transmission %
3.45µm NBP lter
Polyethylene transmission
Resulting transmission
Figure 8. Application of a narrow bandpass (NBP) lter to achieve nearly complete absorption and high emittance (green curve) from polyethylene lm, allowing its temperature measurement.
40
Appendix A
Glossary
absorption (absorption factor). The
amount of radiation absorbed by an
object relative to the received radiation. A number between 0 and 1.
ambient. Objects and gases that emit
radiation towards the object being measured.
atmosphere. The gases between the
object being measured and the camera,
normally air.
autoadjust. A function making a camera
perform an internal image correction.
autopalette. The IR image is shown with
an uneven spread of colors, displaying
cold objects as well as hot ones at the same time.
blackbody. Totally non-reective object. All its radiation is due to its own
temperature.
blackbody radiator. An IR radiating device with blackbody properties used to
calibrate IR cameras.
calculated atmospheric transmission. A transmission value computed from the temperature, the relative humidity of the air, and the distance to the object.
cavity radiator. A bottle shaped radiator with an absorbing inside, viewed through
the bottleneck.
color temperature. The temperature for which the color of a blackbody matches a
specic color.
conduction. The process that makes heat spread into a material.
continuous adjust. A function that
adjusts the image. The function works
all the time, continuously adjusting
brightness and contrast according to the image content.
convection. The process that makes hot air or liquid rise.
dierence temperature. A value that is
the result of a subtraction between two
temperature values.
dual isotherm. An isotherm with two
color bands instead of one.
emissivity (emissivity factor). The amount of radiation coming from an object compared to that of a blackbody.
A number between 0 and 1.
emittance. Amount of energy emitted
from an object per unit of time and area
(W/m).
estimated atmospheric transmission.
A transmission value, supplied by a user,
replacing a calculated one.
external optics. Extra lenses, lters, heat
shields etc. that can be put between the camera and the object being measured.
lter. A material transparent only to some of the infrared wavelengths.
FOV. Field of view: The horizontal angle that can be viewed through an IR lens.
FPA . Focal plane array: A type of IR
detector.
graybody. An object that emits a xed
fraction of the amount of energy of a
blackbody for each wavelength.
IFOV. Instantaneous Field Of View: A
measure of the geometrical resolution of an IR camera.
41
Glossary
image correction (internal or external).
A way of compensating for sensitivity dierences in various parts of live images and also of stabilizing the camera.
infrared. Non-visible radiation, with a wavelength from about 2–13 µm.
IR. Infrared.
isotherm. A function highlighting those
parts of an image that fall above, below,
or between one or more temperature
intervals.
isothermal cavity. A bottle-shaped
radiator with a uniform temperature
viewed through the bottleneck.
Laser LocatIR. An electrically powered
light source on the camera that emits
laser radiation in a thin, concentrated
beam to point at certain parts of the object in front of the camera.
laser pointer. An electrically powered
light source on the camera that emits
laser radiation in a thin, concentrated
beam to point at certain parts of the object in front of the camera.
level. The center value of the temperature scale, usually expressed as a signal value.
manual adjust. A way to adjust the
image by manually changing certain parameters.
NETD. Noise equivalent temperature dierence: A measure of the image noise level of an IR camera.
noise. Undesired small disturbance in the
infrared image.
object parameters. A set of values
describing the circumstances under
which the measurement of an object
was made and the object itself (such as emissivity, ambient temperature, distance, etc.)
object signal. A non-calibrated value
related to the amount of radiation
received by the camera from the object.
palette. The set of colors used to display an IR image.
pixel. A picture element. One single spot
in an image.
radiance. Amount of energy emitted from an object per unit of time, area, and angle (W/m/sr).
radiant power. Amount of energy
emitted from an object per unit of
time (W).
radiation. The process by which electromagnetic energy is emitted by an object or a gas.
radiator. A piece of IR radiating
equipment.
range. The current overall temperature
measurement limitation of an IR camera.
Cameras can have several ranges,
which are expressed as two blackbody temperatures that limit the current calibration.
reference temperature. A temperature which the ordinary measured values can
be compared with.
reection. The amount of radiation
reected by an object relative to the received radiation. A number between 0 and 1.
42
Appendix A
relative humidity. Percentage of water
in the air relative to what is physically possible. Air temperature dependent.
saturation color. The areas that contain
temperatures outside the present level/
span settings are colored with the saturation colors. The saturation colors
contain an “overow” color and an
“underow” color. There is also a third red
saturation color that marks everything
saturated by the detector indicating that the range should probably be changed.
span. The interval of the temperature scale, usually expressed as a signal value.
spectral (radiant) emittance. Amount of
energy emitted from an object per unit of
time, area, and wavelength (W/m/µm).
temperature range. The current overall
temperature measurement limitation of
an IR camera. Cameras can have several
ranges. They are expressed as two
blackbody temperatures that limit the current calibration.
temperature scale. The way in which an IR image currently is displayed. Expressed
as two temperature values limiting
the colors.
thermogram. Infrared image.
transmission (or transmittance) factor.
Gases and materials can be more or less
transparent. Transmission is the amount
of IR radiation passing through them. A number between 0 and 1.
transparent isotherm. An isotherm showing a linear spread of color, instead of covering the highlighted parts of
the image.
visual. Refers to the video mode of an IR camera as opposed to the normal,
thermographic mode. When a camera is
in video mode it captures ordinary video images, while thermographic images are
captured when the camera is in IR mode.
43
Thermographic
Measurement Techniques
Thermographic Measurement Techniques
Introduction
An infrared camera measures and images
the emitted infrared radiation from an object. The fact that radiation is a function of object surface temperature makes it possible for the camera to calculate and display this temperature.
However, the radiation measured by the
camera does not only depend on the temperature of the object but is also a
function of the emissivity. Radiation also
originates from the surroundings and is reected in the object. The radiation from the object and the reected radiation will also be inuenced by the absorption of the atmosphere.
To measure temperature accurately, it
is therefore necessary to compensate for the eects of a number of dierent
radiation sources. This is done on-
line automatically by the camera. The
following object parameters must, however, be supplied for the camera:
The reected temperature The distance between the object and
the camera
The relative humidity• The emissivity of the object
The most important object parameter to
set correctly is the emissivity which, in short, is a measure of how much radiation is emitted from the object, compared to
that from a perfect blackbody.
Normally, object materials and surface treatments exhibit emissivity ranging from approximately 0.1 to 0.95. A highly
polished (mirror) surface falls below 0.1, while an oxidized or painted surface has much higher emissivity. Oil-based paint, regardless of color in the visible spectrum,has an emissivity of over 0.9
in the infrared. Human skin exhibits an
emissivity close to 1. Non-oxidized metals represent an
extreme case of almost perfect opacity
and high spectral reexivity, which does not vary greatly with wavelength. Consequently, the emissivity of metals is low – only increasing with temperature. For non-metals, emissivity tends to be
high and decreases with temperature.
Finding the Emissivity of an Object Using a Thermocouple
Select a reference point and measure its temperature using a thermocouple.
Alter the emissivity until the temperature
measured by the camera agrees with the thermocouple reading. This is
the emissivity value of the reference object. However, the temperature of the
reference object must not be too close to the ambient temperature for this to work.
Finding the Emissivity of an Object Using Reference Emissivity
A tape or paint of a known emissivity
should be put onto the object. Measure the temperature of the tape/paint using
the camera, setting emissivity to the correct value. Note the temperature. Alter the emissivity until the area with the unknown emissivity adjacent to the tape/
paint has the same temperature reading.
The emissivity value can now be read.
The temperature of the reference object must not be too close to the ambient temperature for this to work either.
Appendix B
44
Appendix B
Reected Temperature Parameter
This parameter is used to compensate for the radiation reected in the object and the radiation emitted from the atmosphere between the camera and the object.
If the emissivity is low, the distance very long, and the object temperature relatively close to that of the reected object, it is important to
set and compensate for the reected temperature correctly.
Distance Parameter
This is the distance between the object and the front lens of the camera.
This parameter is used to compensate for the fact that radiation is being absorbed between the object and the camera and the fact that transmittance drops with distance.
Relative Humidity Parameter
The camera can also compensate for the fact that transmittance is somewhat
dependent on the relative humidity of the atmosphere. To do this, set the relative humidity to the correct value. For short distances and normal humidity, the relative humidity can normally be left at a default value of 50%.
Other Parameters
In addition, some cameras and analysis
programs from FLIR Systems allow you to compensate for the following parameters:
Atmospheric temperature – the
temperature of the atmosphere between the camera and the target
External optics temperature – the
temperature of any external lenses or windows used in front of the camera
External optics transmission – the
transmission of any external lenses or windows used in front of the camera
45
History and Theory of
Infrared Technology
History and Theory of Infrared Technology
Less than 200 years ago the existence
of the infrared portion of the
electromagnetic spectrum wasn’t even suspected. The original signicance of the infrared spectrum, or simply “the infrared” as it is often called, as a form of heat radiation is perhaps less obvious
today than it was at the time of its
discovery by Sir William Herschel in 1800 (Figure 1).
Figure 1. Sir William Herschel (1738–1822)
The discovery was made accidentally
during the search for a new optical
material. Sir William Herschel – Royal Astronomer to King George III of England, and already famous for his discovery of the planet Uranus – was searching for an optical lter material to reduce the brightness of the sun’s image in telescopes during solar observations.
While testing dierent samples of colored
glass which gave similar reductions in brightness, he was intrigued to nd that some of the samples passed very little of the sun’s heat, while others passed so
much heat that he risked eye damage
after only a few seconds’ observation.
Herschel was soon convinced of the
necessity of setting up a systematic
experiment with the objective of nding a single material that would give the
desired reduction in brightness as well as the maximum reduction in heat. He began the experiment by actually
repeating Newton’s prism experiment,
but looking for the heating eect rather
than the visual distribution of intensity in the spectrum. He rst blackened the bulb of a sensitive mercury-in-glass thermometer with ink, and with this as
his radiation detector he proceeded to
test the heating eect of the various
colors of the spectrum formed on the top of a table by passing sunlight through a
glass prism. Other thermometers, placed outside the sun’s rays, served as controls.
As the blackened thermometer was moved slowly along the colors of the spectrum, the temperature readings
showed a steady increase from the
violet end to the red end. This was not entirely unexpected, since the Italian researcher, Landriani (Figure 2), in a similar experiment in 1777, had observed much the same eect. It was Herschel, however, who was the rst to recognize
that there must be a point where the
heating eect reaches a maximum, and that measurements conned to the visible portion of the spectrum failed to
locate this point.
Figure 2. Marsilio Landriani (1746–1815)
Appendix C
46
Appendix C
Moving the thermometer into the
dark region beyond the red end of the
spectrum, Herschel conrmed that
the heating continued to increase. The
maximum point, when he found it, lay well beyond the red end – in what is known today as the infrared wavelengths.
When Herschel revealed his discovery,
he referred to this new portion of the electromagnetic spectrum as the thermometrical spectrum. The radiation itself he sometimes referred
to as dark heat, or simply the invisible rays. Ironically, and contrary to popular opinion, it wasn’t Herschel who
originated the term infrared. The word only began to appear in print around
75 years later and it is still unclear who should receive credit as the originator.
Herschel’s use of glass in the prism of his
original experiment led to some early
controversies with his contemporaries
about the actual existence of the infrared
wavelengths. Dierent investigators, in attempting to conrm his work, used various types of glass indiscriminately, having dierent transparencies in the infrared. Through his later experiments,
Herschel was aware of the limited
transparency of glass to the newly­discovered thermal radiation, and he was
forced to conclude that optics for the infrared would probably be doomed to
the use of reective elements exclusively (plane and curved mirrors). Fortunately, this proved to be true only until 1830, when the Italian investigator Melloni (Figure 3) made his great discovery that naturally occurring rock salt (NaCl) – which was available in large enough
natural crystals to be made into lenses
and prisms – is remarkably transparent
to the infrared. The result was that rock salt became the principal infrared optical material and remained so for the next hundred years until the art of synthetic crystal growing was mastered
in the 1930s.
Figure 3. Macedonio Melloni (1798–1854)
Thermometers, as radiation detectors, remained unchallenged until 1829, the year Nobili invented the thermocouple. (Herschel’s own thermometer could be read to 0.2°C (0.036°F), and later models were able to be read to 0.05°C (0.09°F).)
Then a breakthrough occurred; Melloni connected a number of thermocouples
in series to form the rst thermopile. The new device was at least 40 times as sensitive as the best thermometer of the day for detecting heat radiation – capable
of detecting the heat from a person standing three meters away.
The rst so-called heat-picture became possible in 1840, the result of work by Sir John Herschel, son of the discoverer of
infrared and a famous astronomer in his own right. Based upon the dierential
evaporation of a thin lm of oil when
exposed to a heat pattern focused upon
it, the thermal image could be seen by
reected light where the interference
eects of the oil lm made the image
47
History and Theory of Infrared Technology
visible to the eye. Sir John also managed to obtain a primitive record of the thermal image on paper, which he called
a thermograph.
Figure 4. Samuel P. Langley (1834–1906)
The improvement of infrared-detector sensitivity progressed slowly. Another major breakthrough, made by Langley (Figure 4) in 1880, was the invention
of the bolometer. This consisted of a thin blackened strip of platinum connected in one arm of a Wheatstone bridge circuit upon which the infrared radiation was focused and to which a
sensitive galvanometer responded. This instrument is said to have been able to
detect the heat from a cow at a distance
of 400 meters.
An English scientist, Sir James Dewar, rst introduced the use of liqueed gases as cooling agents (such as liquid nitrogen with a temperature of –196°C (–320.8°F)) in low temperature research. In 1892 he invented a unique vacuum insulating
container in which it is possible to store
liqueed gases for entire days. The common thermos bottle, used for storing hot and cold drinks, is based upon his invention.
Between the years 1900 and 1920, the inventors of the world “discovered”
infrared. Many patents were issued for
devices to detect personnel, artillery, aircraft, ships – and even icebergs. The rst operating systems, in the modern sense, began to be developed during World War I, when both sides had research programs devoted to the
military exploitation of the infrared. These programs included experimental
systems for enemy intrusion/detection, remote temperature sensing, secure communications, and “ying torpedo” guidance. An infrared search system
tested during this period was able to detect an approaching airplane at a
distance of 1.5 km (0.94 miles) or a person more than 300 meters (984 ft.) away.
The most sensitive systems up to this time were all based upon variations of the bolometer idea, but the period between the two world wars saw the development of two revolutionary new infrared detectors: the image converter and the photon detector. At rst, the image converter received the greatest attention by the military, because it enabled an observer for the rst time in history to literally see in the dark. However, the sensitivity of the image converter was limited to the near infrared wavelengths,
and the most interesting military targets
(enemy soldiers) had to be illuminated by infrared search beams. Since this involved the risk of giving away the observer’s position to a similarly-equipped enemy observer, it is understandable that military interest in the image converter eventually faded.
The tactical military disadvantages of so-called active (search beam
48
Appendix C
equipped) thermal imaging systems provided impetus following World War II for extensive secret military infrared-
research programs into the possibilities
of developing passive (no search beam) systems around the extremely sensitive photon detector. During this period,
military secrecy regulations completely
prevented disclosure of the status of infrared-imaging technology. This
secrecy only began to be lifted in the
middle of the 1950s and from that time adequate thermal-imaging devices nally began to be available to civilian science
and industry.
Theory of Thermography
Introduction
The subjects of infrared radiation and the related technique of thermography are still new to many who will use an infrared
camera. In this section, the theory behind thermography will be given.
The Electromagnetic Spectrum
The electromagnetic spectrum (Figure
5) is divided arbitrarily into a number of wavelength regions, called bands,
distinguished by the methods used to produce and detect the radiation. There is no fundamental dierence between radiation in the dierent bands of the electromagnetic spectrum. They are all
governed by the same laws and the only
dierences are those due to dierences
in wavelength.
Thermography makes use of the infrared
spectral band. At the short-wavelength
end of the spectrum the boundary lies at
the limit of visual perception, in the deep red. At the long-wavelength end of the spectrum it merges with the microwave radio wavelengths in the millimeter
range.
The infrared band is often further
subdivided into four smaller bands, the
boundaries of which are also arbitrarily
Figure 5. The electromagnetic spectrum
10 nm 100 nm 1 µm 10 µm 100 µm 1 mm 10 mm 100 mm 1 m 10 m 100 m 1 km
1 X-ray 2 UV 3 Visible 4 IR 5 Microwaves 6 Radio waves
2 µm
1234 56
13 µm
49
History and Theory of Infrared Technology
chosen. They include: the near infrared
(0.75–3 µm), the middle infrared (3–6 µm), the far infrared (6–15 µm), and the extreme infrared (15–100 µm). Although the wavelengths are given in µm (micrometers), other units are often used to measure wavelength in this spectral region, for exmple, nanometer (nm) and Ångström (Å).
The relationships between the dierent
wavelength measurements is:
10,000 Å = 1,000 nm = 1 µm
Blackbody Radiation
A blackbody is dened as an object
that absorbs all radiation that impinges
on it at any wavelength. The apparent
misnomer black relating to an object emitting radiation is explained by
Kirchho’s Law (after Gustav Robert Kirchho, shown in Figure 6), which
states that a body capable of absorbing
all radiation at any wavelength is equally
capable in the emission of radiation.
Figure 6. Gustav Robert Kirchho (1824–1887)
The construction of a blackbody source
is, in principle, very simple. The radiation
characteristics of an aperture in an
isothermal cavity made of an opaque
absorbing material represents almost
exactly the properties of a blackbody. A
practical application of the principle to the construction of a perfect absorber of radiation consists of a box that is light tight except for an aperture in one of
the sides. Any radiation that then enters
the hole is scattered and absorbed
by repeated reections, so only an innitesimal fraction can possibly escape.
The blackness which is obtained at the aperture is nearly equal to a blackbody
and almost perfect for all wavelengths.
By providing such an isothermal cavity with a suitable heater, it becomes what is termed a cavity radiator. An isothermal cavity heated to a uniform temperature generates blackbody radiation, the
characteristics of which are determined
solely by the temperature of the cavity. Such cavity radiators are commonly used
as sources of radiation in temperature reference standards in the laboratory for
calibrating thermographic instruments, such as a FLIR Systems camera, for
example.
If the temperature of blackbody radiation
increases to more than 525°C (977°F), the source begins to be visible so that it no
longer appears black to the eye. This is the incipient red heat temperature of the
radiator, which then becomes orange
or yellow as the temperature increases
further. In fact, the denition of the so-
called color temperature of an object is the temperature to which a blackbody
would have to be heated to have the
same appearance.
Now consider three expressions that
describe the radiation emitted from a blackbody.
50
Appendix C
Planck’s Law
Figure 7. Max Planck (1858–1947)
Max Planck (Figure 7) was able to describe
the spectral distribution of the radiation from a blackbody by means of the following formula:
πhc3 Wλb =
_______________
× 10–6 [Watt/m µ m ]
λ
(ehc/λkT
– 1)
where:
Wλb Blackbody spectral radiant emittance
at wavelength
λb
c Velocity of light = 3 × 108 m/s h Planck’s constant = 6.6 × 10
–34
Joule
sec
k Boltzmann’s constant = 1.4 × 10
–23
Joule/K
T Absolute temperature (K) of a
blackbody
λ Wavelength (µm)
The factor 10–6 is used since spectral emittance in the curves is expressed in
Watt/mm. If the factor is excluded, the dimension will be Watt/mµm.
Planck’s formula, when plotted graphically for various temperatures, produces a family of curves. Following
any particular Planck curve, the spectral emittance is zero at λ = 0, then increases rapidly to a maximum at a wavelength µmax and after passing it approaches zero again at very long wavelengths. The higher the temperature, the shorter the wavelength at which maximum occurs. See Figure 8.
Wien’s Displacement Law
By dierentiating Planck’s formula with respect to λ and nding the maximum, we have:
2898
λ
max
=
______
[µm]
T
This is Wien’s formula (after Wilhelm Wien, shown in Figure 9), which expresses mathematically the common observation that colors vary from red to orange or
yellow as the temperature of a thermal
radiator increases (Figure 10). The wavelength of the color is the same as
8
7
6
5
4
3
2
1
0
0 2 4 6 8 10 12 14
900 K
800 K
700 K
600 K
500 K
Spectral radiant emittance [W/cm
2
× 10
3
(µm)]
Wa velength (µm)
Figure 8. Blackbody spectral radiant emittance according to Planck’s law, plotted for various absolute temperatures.
51
History and Theory of Infrared Technology
the wavelength calculated for λ
max
. A good approximation of the value of µmax for a given blackbody temperature is obtained by applying the rule-of-thumb 3,000/T µm. Thus, a very hot star such as Sirius (11,000 K), emitting bluish-white
light, radiates with the peak of spectral
radiant emittance occurring within
the invisible ultraviolet spectrum at wavelength 0.27 µm.
The sun (approx. 6,000 K) emits yellow light, peaking at about 0.5 µm in the middle of the visible light spectrum.
At room temperature (300 K) the peak of radiant emittance lies at 9.7 µm in the far infrared, while at the temperature of liquid nitrogen (77 K) the maximum of the almost insignicant amount of radiant emittance occurs at 38 µm in the extreme infrared wavelengths.
By integrating Planck’s formula from λ = 0 to λ = ∞, we obtain the total radiant emittance (Wb) of a blackbody:
Wb = σT4 [Watt/m] This is the Stefan-Boltzmann formula
(after Josef Stefan and Ludwig Boltzmann, shown in Figure 11), which states that the total emissive power of a
blackbody is proportional to the fourth power of its absolute temperature.
Graphically, Wb represents the area below the Planck curve for a particular
temperature. It can show that the radiant emittance in the interval λ = 0 to λ
max
is
only 25% of the total, which represents
Figure 9. Wilhelm Wien (1864–1928)
10
5
10
4
10
3
10
2
10
1
10
0
10
–1
0 5 10 15 20 25 30
1000 K
900
800
700
600
500
400
300
200
100
Spectral radiant emittance [W/cm
2
(µm)]
Wa velength (µm)
Figure 10. Planckian curves plotted on semi­log scales from 100 K to 1000 K. The dotted line represents the locus of maximum radiant emittance at each temperature as described by Wien’s displacement law.
Figure 11. Josef Stefan (1835–1893) and Ludwig Boltzmann (1844–1906)
52
Appendix C
approximately the amount of the sun’s radiation that lies inside the visible light
spectrum.
Using the Stefan-Boltzmann formula
to calculate the power radiated by the
human body, at a temperature of 300 K
and an external surface area of approx.  m, we obtain 1 kW. This power loss could not be sustained if it were not for the compensating absorption of radiation from surrounding surfaces at
room temperatures, which do not vary
too drastically from the temperature of
the body – or, of course, the addition of
clothing.
Non-blackbody Emitters
So far, only blackbody radiators and blackbody radiation have been discussed. However, real objects almost never comply with these laws over an extended wavelength region – although they may approach the blackbody behavior in certain spectral intervals. For example, a
certain type of white paint may appear
perfectly white in the visible light spectrum, but becomes distinctly gray at about 2 µm, and beyond 3 µm it is
almost black.
There are three processes that can
prevent a real object from acting like
a blackbody: a fraction of the incident
radiation α may be absorbed, a fraction ρ may be reected, and a fraction τ may be
transmitted. Since all of these factors are
more or less wavelength dependent, the
subscript λ is used to imply the spectral
dependence of their denitions. Thus:
The spectral absorptance α
λ
= the ratio
of the spectral radiant power absorbed by an object to that incident upon it.
The spectral reectance ρ
λ
= the ratio
of the spectral radiant power reected by an object to that incident upon it.
The spectral transmittance τ
λ
= the
ratio of the spectral radiant power transmitted through an object to that incident upon it.
The sum of these three factors must always add up to the whole at any
wavelength, so we have the relation: αλ + ρλ + τλ = 1 For opaque materials τλ = 0 and the
relation simplies to: αλ + ρλ = 1 Another factor, called the emissivity, is
required to describe the fraction ε of the radiant emittance of a blackbody
produced by an object at a specic temperature. Thus, we have the denition: spectral emissivity ελ = the
ratio of the spectral radiant power from an object to that from a blackbody at the
same temperature and wavelength. Expressed mathematically, this can
be written as the ratio of the spectral emittance of the object to that of a blackbody as follows:
Wλ0 ελ =
_____
W
λb
Generally speaking, there are three types of radiation sources, distinguished by the
ways in which the spectral emittance of
each varies with wavelength (Figures 12 and 13).
A blackbody, for which ελ = ε = 1 A graybody, for which ελ = ε = constant
less than 1 A selective radiator, for which ε varies
with wavelength
53
History and Theory of Infrared Technology
According to Kirchho’s law, for any material the spectral emissivity and
spectral absorptance of a body are
equal at any specied temperature and wavelength. That is:
ελ = α
λ
From this we obtain, for an opaque material (since αλ + ρλ = 1):
ελ + ρλ = 1
For highly polished materials ελ
approaches zero, so for a perfectly reecting material (for example. a perfect mirror) we have:
ρλ = 1
For a graybody radiator, the Stefan­Boltzmann formula becomes:
W = εσT4 [Watt/m]
This states that the total emissive power
of a graybody is the same as a blackbody at the same temperature reduced in proportion to the value of ε from the graybody.
Blackbody
Selective radiator
Graybody
Spectral radiant emittance
Wa velength (µm)
Figure 12. Spectral radiant emittance of three types of radiators.
Blackbody
Selective radiator
Graybody
Spectral emissivity
Wa velength
1.0
0.5
0.0
Figure 13. Spectral emissivity of three types of radiators.
Infrared Semi-transparent Materials
Consider a non-metallic, semi-
transparent body in the form of a thick at plate of plastic material. When the
plate is heated, radiation generated within its volume must work its way
toward the surfaces through the material in which it is partially absorbed.
Moreover, when it arrives at the surface,
some of it is reected back into the
interior. The back-reected radiation is again partially absorbed, but some of it arrives at the other surface, through which most of it escapes, but part of it is reected back again. Although the progressive reections become weaker and weaker, they must all be added up
when the total emittance of the plate is sought. When the resulting geometrical
series is summed, the eective emissivity of a semi-transparent plate is obtained as:
(1 –ρλ) (1 – τλ)
ελ =
______________
1 – pλt
λ
When the plate becomes opaque this formula is reduced to the single formula:
ελ = 1 – p
λ
54
Appendix C
This last relation is a particularly
convenient one, because it is often easier
to measure reectance than to measure
emissivity directly.
The Measurement Formula
As already mentioned, when viewing an object, the camera receives radiation not only from the object itself, it also
collects radiation from the surroundings
reected via the object surface. Both
these radiation contributions become attenuated to some extent by the atmosphere in the measurement path. To this comes a third radiation contribution from the atmosphere itself.
This description of the measurement
situation, as illustrated in Figure 14, is so
far a fairly true description of the real conditions. What has been neglected could for instance be sun light scattering in the atmosphere or stray radiation from
intense radiation sources outside the eld of view. Such disturbances are dicult to quantify. However, in most cases
they are fortunately small enough to be
neglected. In case they are not negligible, the measurement conguration is likely
to be such that the risk for disturbance
is obvious, at least to a trained operator.
It is then his responsibility to modify
the measurement situation to avoid the disturbance (by changing the viewing direction, shielding o intense radiation sources, etc.).
Accepting the description above, we can use Figure 14 to derive a formula for the
calculation of the object temperature from the calibrated camera output.
Assume that the received radiation
power W from a blackbody source of temperature T
source
on short distances
generates a camera output signal U
source
that is proportional to the power input
(power linear camera). We can then write (Equation 1):
U
source
= CW (T
source
) or, with simplied notation: U
source
= CW
source
where C is a constant.
Figure 14. A schematic representation of the general thermographic measurement situation.
ε τ W
obj
ε W
obj
Surroundings
Object
Atmosphere
Camera
ε
re
= 1
ε
T
obj
(1 ε) τ W
re
(1 ε) W
re
τ
(1 τ) W
atm
T
atm
T
re
W
re
55
History and Theory of Infrared Technology
Should the source be a graybody with
emittance ε, the received radiation would
consequently be εW
source
.
We are now ready to write the three collected radiation power terms:
Emission from the object1. = ετW
obj
, where
ε is the emittance of the object and τ is
the transmittance of the atmosphere. The object temperature is T
obj
.
Reected emission from ambient sources2.
= (1 – ε)τW
re
, where (1 – ε) is the
reectance of the object. The ambient
sources have the temperature T
re
.
It has here been assumed that the temperature T
re
is the same for all
emitting surfaces within the half-
sphere seen from a point on the object surface. This is of course sometimes a
simplication of the true situation. It is, however, a necessary simplication in order to derive a workable formula,
and T
re
can – at least theoretically – be
given a value that represents an ecient
temperature of a complex surrounding.
Note also that we have assumed that the emittance for the surroundings = 1. This is correct in accordance with Kirchho’s law: All radiation impinging on the surrounding surfaces will eventually
be absorbed by the same surfaces.
Thus the emittance = 1. (Note, though,
that the latest discussion requires the complete sphere around the object to be
considered.)
Emission from the atmosphere3. = (1 – τ)τW
atm
, where (1 – τ) is the
emittance of the atmosphere. The temperature of the atmosphere is T
atm
.
The total received radiation power can now be written (Equation 2):
W
tot
= ετW
obj
+ (1 – ε)τW
re
+ (1 – τ)W
atm
We multiply each term by the constant
C of Equation 1 and replace the CW products by the corresponding U according to the same equation, and get (Equation 3):
U
tot
= ετU
obj
+ (1 – ε)τU
re
+ (1 – τ)U
atm
Solve Equation 3 for U
obj
(Equation 4):
1 1 – ε 1 – τ U
obj
=
___
U
tot
_____
U
re
_____
U
atm
ετ ε ετ
This is the general measurement formula used in all the FLIR Systems
thermographic equipment. The voltages of the formula are given in Table 1.
Table 1. Voltages
U
obj
Calculated camera output voltage for a
blackbody of temperature T
obj
, which is a
voltage that can be directly converted into
true requested object temperature.
U
tot
Measured camera output voltage for the
actual case.
U
re
Theoretical camera output voltage for a
blackbody of temperature T
re
according to
the calibration.
U
atm
Theoretical camera output voltage for a
blackbody of temperature T
atm
according to
the calibration.
The operator has to supply a number of
parameter values for the calculation:
object emittance ε• relative humidity
T
atm
object distance (D
obj
)
(eective) temperature of the object
surroundings or the reected ambient temperature T
re
the temperature of the atmosphere T
atm
56
Appendix C
This task can sometimes be a heavy
burden for the operator since there are
normally no easy ways to nd accurate values of emittance and atmospheric
transmittance for the actual case. The two temperatures are normally less of a
problem provided the surroundings do
not contain large and intense radiation sources.
A natural question in this connection is:
How important is it to know the right
values of these parameters? It could
be of interest to get a feeling for this problem by looking into some dierent measurement cases and compare the
relative magnitudes of the three radiation terms. This will give indications about when it is important to use correct values
for which parameters.
Figures 15 and 16 illustrate the relative
magnitudes of the three radiation contributions for three dierent object
temperatures, two emittances, and two
spectral ranges: SW and LW. Remaining
parameters have the following xed values:
τ = 0.88
T
re
= +20°C (+68°F)
T
atm
= +20°C (+68°F)
It is obvious that measurement of low
object temperatures is more critical than measuring high temperatures since the “disturbing” radiation sources are
relatively much stronger in the rst case. Should the object emittance be low, the situation would be more dicult.
We have nally to answer a question
about the importance of being allowed
to use the calibration curve above the highest calibration point, what we call
extrapolation. Imagine that we in a
certain case measure U
tot
= 4.5 volts. The
highest calibration point for the camera
was in the order of 4.1 volts, a value unknown to the operator. Thus, even if
the object happened to be a blackbody
(U
obj
= U
tot
), we are actually performing
extrapolation of the calibration
curve when converting 4.5 volts into
temperature.
0°C (3 2°F)
0.6
0.8
20 °C ( 68°F)
Object T e mperat ure
Emittance
50 °C ( 1 22 °F)
Object Radia tion
Reect ed Radia tion
At mosp here Rad iation
Figure 15. Relative magnitudes of radiation sources under varying measurement conditions (SW camera). Fixed parameters: τ = 0.88; T
re
=
20°C (+68°F); T
atm
= 20°C (+68°F).
0°C (3 2°F)
0.6
0.8
20 °C ( 68°F)
Object T e mperat ure
Emittance
50 °C ( 1 22 °F)
Object Radia tion
Reect ed Radia tion
At mosp here Rad iation
Figure 16. Relative magnitudes of radiation sources under varying measurement conditions (LW camera). Fixed parameters: τ = 0.88; T
re
=
20°C (+68°F); T
atm
= 20°C (+68°F).
57
History and Theory of Infrared Technology
Let us now assume that the object is
not black, it has an emittance of 0.75 and the transmittance is 0.92. We also
assume that the two second terms of
Equation 4 amount to 0.5 volts together. Computation of U
obj
by means of
Equation 4 then results in U
obj
= 4.5 /
0.75 / 0.92 – 0.5 = 6.0. This is a rather extreme extrapolation, particularly when considering that the video amplier might limit the output to 5 volts. Note, though, that the application
of the calibration curve is a theoretical
procedure where no electronic or other limitations exist. We trust that if there had been no signal limitations in the
camera, and if it had been calibrated far beyond 5 volts, the resulting curve would have been very much the same as our real curve extrapolated beyond 4.1 volts, provided the calibration algorithm is based on radiation physics, like the FLIR Systems algorithm. Of course there must
be a limit to such extrapolations.
58
Appendix D
Command Syntax
Examples for A320 Resource Socket Services
Physical Interface
From the camera, there is only one physical interface for data transfer, Ethernet. Analog video also exists, but
is not considered a data transfer interface.
Low Level Protocols
For Ethernet, only TCP/IP is supported.
The camera should seamlessly work on
any LAN, provided that a proper IP adress, netmask, and possibly gateway is set in
the camera.
No FLIR specic device drivers are required, so any type of computer and
operating systems supporting TCP/IP should work.
Functionality
The two main ways of controlling the camera are through the command control interface and through the resource control interface. Image
streaming, le transfer, and other functionality is provided through the IP services interface.
Command Control
Commands can be given to the
“commandshell” in the camera. Some commands are “standardcommands” like “dir” and “cd” that operate directly
on the camera themselves. Others rely on camera resources (see below), for example the “level” command.
It is possible to run independent instances of the command shell with the
“telnet” service on established TCP/IP
connections.
Resource Control
Most, but not all, software functionality
is exposed through software resources. Those that are familiar with the Microsoft
Windows registry will recognize this concept. However, in the camera a
resource node can also represent a
software function that, upon read or write, actively interacts with the
camera software.
The following lists some introductory details about software resources and resource nodes.
Resources are organized in a hierarchy • (like in a tree).
Resource nodes can be data holders (of • for example, calibration data).
Resource nodes can be connected • to hardware (for example, to internal temperature sensor values).
Resource nodes can be connected to • software (for example, to spotmeter values)
Resource nodes have a type (double, • int32, ascii, etc.) and certain attributes (readonly, read/write).
Resources can be reached through • commands like “rls” and “rset,” and
through the IP FLIR Resource socket
service.
IP Services
It is possible to access the system independent of physical interface using
TCP/IP with exposed services such as telnet, ftp, http, CIFS, FLIR resource socket, and FLIR RTP. More than one service and possibly more than one
59
Command Syntax Examples for
A320 Resource Socket Services
instance of the service can be run
simultaneously.
telnet
Command control, mainly for manual
typing. Typical clients are the standard telnet command on a PC or teraterm.
FTP
File transfer to/from the camera using FTP client software on a PC. Typical clients are the standard FTP command on a PC or WS_FTP.
http
Web server. Typical clients are Microsoft
Internet Explorer and Firefox.
CIFS
PC network le access service. This service makes it possible to map a drive on the PC to the camera le system. The
intended client software is built into all
relevant Windows versions. By default,
only the image folders are accessible in
this way. To access the whole ash le system, mount the xxx.xxx.xxx.xxx/root$ drive.
FLIR Resource Socket
It is possible to directly read and write nodes of the software resource tree from
a PC. Standard sockets are used, but there are no standard client programs available.
Image Streaming
Set-up
The available streams are described and presented using SDP (Session Description Language, RFC 2327). The SDP content is accessed using RTSP (Real Time Streaming Protocol, RFC 2326).
The RTSP/DESCRIBE command lists the
available streams.
Table 1
MPEG4 Compressed video in three sizes
(640x480, 320x240, 160x128) FCAM FLIR usage only. Raw IR-signal or temperature linear IR-image
(two types) in two sizes (320×240,
160×120). The pixels are transferred in
network byte order (big endian).
The RTSP/SETUP command establishes an RTP-based transport session using one of
the formats.
The RTSP/GET_PARAMETER command
gets the current framerate and format.
The RTSP/SET_PARAMETER command
sets the current framerate and format.
The RTSP/PLAY and RTSP/PAUSE
commands control the image stream.
The RTSP/TEARDOWN command closes
the transport session.
MPEG4
The MPEG4 streams use RTP/UDP/IP for transport (RTP = Real time Transport Protocol, RFC 1889). The MPEG bit stream is packetized according to RFC 3016.
On the receiver side, FLIR supplies a DirectShow component (Win32, PC platform) which is able to receive the MPEG4 bit stream. The component is able to receive the MPEG4 bit stream according to RFC 3016. The bit stream is reassembled and forwarded as video samples of FOURCC type MP4V. Several MPEG4 decoders can be used, for example from 3ivx ($7 per license) or a free decoder (dshow). The DirectShow
component can be used by any application that wishes to display the
MPEG4 video stream.
60
Appendix D
IR Streams
The raw IR streams use RTP/UDP/IP
for transport. The transport format is
according to RFC 4175 (RTP Payload Format for Uncompressed Video).
The raw image frame rates are up to
about 7.8 Hz.
These raw formats are provided:
Table 2.
0 16-bit uncompressed IR image linear in signal 1 16-bit uncompressed IR image linear in
temperature, resolution 0.1 K (range 0–6553 K)
16-bit uncompressed IR image linear in
temperature, resolution 0.01 K (range 0–655 K)
On the client side, FLIR supplies a DirectShow component (Win32, PC platform) which is able to setup and receive the IR streams. The IR stream is
reassembled and forwarded as samples
of FOURCC type Y160 (this FOURCC type is only preliminary at this stage). The
DirectShow component can be used by any application that wishes to display the IR stream or want to grab samples from the stream.
DHCP
The camera supports the client part of
the Dynamic Host Conguration Protocol (DHCP).
Remote Detection
Multicast DNS (Bonjour)
To query Bonjour for local FLIR IR
cameras, use:
Service name: ir-ircam
Protocol type: _tcp Domain: local
Table 3: TXT records
Name Value Explanation
Model S ID A320 Camera ID
GID Gen_A Generic ID
SI FFF_RTSP Streaming Interface SIV 1.0.0 Streaming interface
version
CI RTREE Command interface CIV 1.0.0 Command interface
version
For more information, see
www.dns-sd.org.
For complete information, see FLIR’s A320
ICD manual.
Quick Summary of FLIR IR Cameras
Photon A320 A325
Sensor Type bolometer bolometer bolometer Pixel Resolution 324×256 320×240 320×240 Pixel Pitch 38μm m m Spectral Ranges 7.5μm – 13.5μm 7.5μm – 13.0μm 7.5μm – 13.0μm
Dynamic Range 14-bit
14-bit, Signal +
Temp Linear
14-bit, GiGE Vision +
Gen<i>Cam
Internal Temperature Calibration
Ambient Drift Compensation Temperature Calibration Imager Only Full Frame Rate 30 fps 30 fps 60 fps
Digital Data Output
GigE (optional)
or Serial
Ethernet GigE
Analog Video RS-170 RS-170
Command and Control
RS-232 or
GigE (optional)
Ethernet GigE
On-Board Analytics Motorized Focus Auto Focus Digital I/O Triggering Options
Remote Camera Control Proprietary
Web, TCP/IP
(Open Protocol)
GeniCam
PoE Compatible Messaging FTP, SMTP SDK Support LabVIEW Compatibility
Aperture
f/1.3, f/1.4, f/1.4, f/1.7
lens dependent
f/1.3 f/1.3
Filtering Options
Available Optics
14.25mm 19mm 35mm 50mm
18mm 30mm 10mm
18mm 30mm 10mm
Appendix E
Published by FLIR Systems Incorporated www.goinfrared.com
Loading...