Cognex Edge Tools User Manual

Download

Page 1

Objectives

At the end of this section, participants will be able to:

 Apply the Property Sheet parameters and auto-inserted

information for the Edge tool to a sample image

Edge Tools

Section 5

In the fifth section of the In-Sight Vision Suite Standard class, we will focus on Edge Tools.

In machine vision terminology, an Edge is defined as the boundary (either a line, arc, or circle) between two adjacent pixel groups with contrasting greyscale values. The In-Sight Edge Tools are used to detect and process statistics about the found edges. In-Sight vision systems perform edge detection and processing using the Edge Tool’s functions.

 Describe the two groups of Edge tool functions

 Explain why the region for the Edge tool must be rotated to

detect a horizontal edge

Version 24.2.1 | 30.August.2024

At the end of this section, participants will be able to:

 Apply the Property Sheet parameters and auto-inserted information for

the Edge tool to a sample image

 Describe the two groups of Edge tool functions  Explain why the region for the Edge tool must be rotated to detect a

horizontal edge

Section 5 | Slide 1

Section 5 | Slide 2

Page 2

Block Inspection

Steps:

1. Determine the presence and position of block.



2. Determine if the block has a gouge.



3. Determine the block length.

4. Determine if the hole sizes are within tolerance.

Edge Functions

Functions that extract edges

Functions that operate on previously extracted edges

Version 24.2.1 | 30.August.2024

Block Inspection Steps:

1. Determine the presence and position of the block using the FindPatterns or PatMax Tool – this was completed in section 3.

2. Determine if the block has a gouge using the ExtractHistogram tool – just completed.

3. Determine if the block length is within tolerance using the FindSegment tool.

4. Determine if the hole sizes are within tolerance using the ExtractBlobs tool.

Version 24.2.1 | 30.August.2024

Extract Edges:

 FindSegment: Locates a pair of Edge within an image region and computes the arc

distance between them. FindSegment forms a one-dimensional projection of the image region by summing pixel values in the x-direction of the region. Edge transitions are extracted from the projected image data. The arc segment over which the edge-to-edge distance is computed is derived from the region used to extract the edges.

 FindLine: Locates a single straight-line edge within an image region. FindLine forms a

one-dimensional projection of the image region by summing pixel values on radial line segments scanned in the positive y-direction relative to the region's local coordinate system. Edge transitions are extracted from the projected image data.

Operate:

 PairDistance: Computes the distance between two edges of an Edge data structure.

PairDistance references two edges, by index number, and reports the distance between their midpoints. The segment over which the edge-to-edge distance is computed is derived from the region originally used to extract the edges from the image.

 PairEdges: Sorts arrays of edges into arrays of edge pairs. PairEdges scans an Edge

data structure for edge pairs that fall within the specified tolerance. Edges belonging to a pair are ordered consecutively, such that the first edge of a pair is on an even index number and the second edge of a pair is on an odd index number.

To view a detailed list of each Edge function, go to the Help File.

Section 5 | Slide 3

Section 5 | Slide 4

Page 3

Edges

Edge Applications

Edges represent a location in an image where a transition from dark to light

(or vice versa) occurs

Edges may be straight, curved, or even a complete circle

Version 24.2.1 | 30.August.2024

In machine vision terminology, an Edge is defined as the boundary (either a line, arc or circle) between two adjacent pixel groups with contrasting greyscale values. The In-Sight Edge Tools are used to detect and process statistics about the found edges.

Edges can be comprised of a single edge, or a pair of edges, which consists of two transitions from dark to light or light to dark as shown above.

Version 24.2.1 | 30.August.2024

This slide shows some of the applications that you can use an Edge tool to complete:

 Gauge parts or part features  Find circle features (center and radius)  Use to quickly locate parts (by finding part edges)  Determine relative contrast using score (0-100)

Section 5 | Slide 5

Section 5 | Slide 6

Page 4

FindSegment

FindSegment: Property Sheet

Version 24.2.1 | 30.August.2024

The FindSegment Tool locates a pair of edges within an image region and computes the arc distance between them. FindSegment forms a one-dimensional projection of the image region by summing pixel values on radial line segments scanned in the positive Y-direction relative to the region’s local coordinate system.

Edge transitions are extracted from the projected image data. The arc segment over which the edge-to-edge distance is computed from the region used to extract the edges.

The direction of the arrow in the search region is the most important point to understand in this section.

The FindSegment Property Sheet includes the following:

 Image – Reference to target image cell  Fixture – Where tool should fixture itself  Region – Region specifying interest zone  Segment Color – Grayscale intensity between edges (black or white)  Find By – Helps select from multiple pairs  Accept Thresh – Minimum contrast score  Normalize Score – Helps find low contrast edges  Angle Range – Allowed angle variation for edges  Edge Width – Number of pixels over transition  Show – Graphic options for display

Version 24.2.1 | 30.August.2024

Section 5 | Slide 7

Section 5 | Slide 8

Page 5

FindSegment: Find By

FindSegment: Region

Pay attention to the

arrow direction!

Version 24.2.1 | 30.August.2024

The Find By parameter specifies the edge selection technique.

The choices are:

 Widest Segment (default) finds the two edges farthest apart from which a

segment of the specified Segment Color can be identified.

 Strongest Edges finds the two edges with the strongest response scores from

which a segment of the specified Segment Color can be identified.

 Widest Contiguous Segment finds the two edges from which a segment of the

specified Segment Color can be identified without any other edges between them.

 Narrowest Segment finds the two edges nearest each other from which a

segment of the specified Segment Color can be identified.

This parameter can help when we have multiple edge pairs candidate in the search region.

Version 24.2.1 | 30.August.2024

The Edge Response Chart (displayed when the Show parameter is set to show all: input, result and chart) displays the first derivative of the greyscale values found in the ROI; with peaks and valleys representing the major edge transitions as show above.

The Accept Thresh/Minimum Contrast parameter is used to set a minimum peak height (contrast threshold). With this set, any peaks that are lower than the minimum peak height are excluded from the results. This allows the inspection analysis to be limited to just those edges in the image that are of a certain magnitude.

Use the function’s edge response chart to determine the correct contrast threshold.

 The Score axis is defined by the Score (100 to -100) and the Accept Threshold

parameter set for the tool. Peaks (or positive scores) indicate that the edge transitions from dark to light, while valleys (or negative scores) indicate that the edge transitions from light to dark (a score of 0 indicates no edge was detected.

 The Offset axis refers to the ROI of detecting the edge feature, where 0

represents the beginning of the region and the right-hand value marks the maximum width (in pixels) of the region. The apex of the peaks or valleys along the Offset axis indicates the position of the found edge within the ROI.

Section 5 | Slide 9

Section 5 | Slide 10

Page 6

FindSegment: Chart

FindSegment: Segment Color

NOT

FOUND

Version 24.2.1 | 30.August.2024

It is important to be aware that Edge tools have a specific direction in which they search – notice the direction of the arrows above. The arrow must touch the edge in order to find it – the arrow in the first example is parallel to the edge so it does not find it.

OTE: The way that you draw your ROI is the direction of the search arrow.

Version 24.2.1 | 30.August.2024

Segment Color – specifies the color of the segment to be located. FindSegment will only report edge pairs of the specified polarity.

 0 = black (default) – Specifies black or white-to-black followed by black-to-white polarity.

 1 = white – Specifies white or black-to-white followed by white-to-black polarity.

Section 5 | Slide 11

Section 5 | Slide 12

Page 7

Edges: Vision Data Access

Functions

Summary

 There are two groups of Edge Tool functions:

 Functions that find edges

 Functions that operate on edges returned by the

functions in the first group

Version 24.2.1 | 30.August.2024

This is an overview of the VDA functions available for Edge tools from the Help file.



GetAngle(Edges, [Index]): Returns the Angle of the indexed edge.



GetContrast(Edges): Returns the average contrast between the foreground and background (in grey levels, 0 to 255) found. Contrast is positive for black-to-white transitions, and negative for white-to-black transitions. The Edge structure must be created by a FindCircleMinMax function.



GetEdgeDistance(Edges, [Index]): Returns the distance between two edges (edge pair). The Edge structure must be created by a Caliper function.



GetMax(Edges): Returns the radius of the maximum deviation from the best edge. The Edge structure must be created by a FindCircleMinMax function.



GetMin(Edges): Returns the radius of the minimum deviation from the best edge. The Edge structure must be created by a FindCircleMinMax function.



GetNFound(Edges): Returns the number of edges found.



GetPosition(Edges, [Index]): Returns the region X position in pixels of the edge or the edge pair center.



GetRadius(Edges, [Index]): Returns the radius of a specified circle or arc.



GetScore(Edges, [Index1], [Index2]): Returns the Score value (0-100) from the indexed edge. Score is positive for black-to-white transitions and negative for white-to-black transitions (except for FindSegment and FindCircleMinMax where Score is always positive).



GetSDev(Edges): Returns the standard deviation value. The Edge structure must be created by a FindCircleMinMax function. Note: This calculation is a biased standard deviation, where the denominator is N-1 (an unbiased standard deviation has a denominator of N).



GetX(Edges, [Index1], [Index2]): Returns a x-coordinate. Index1 selects the edge number, and Index2 specifies an end point (0 = top-most point; 1 = bottom-most point).



GetY(Edges, [Index1], [Index2]): Returns a y-coordinate. Index1 selects the edge number, and Index2 specifies an end point (0 = left-most point; 1 = right-most point).

In this section, we covered the following:

There are two groups of Edge Tool functions:

 Functions that find edges  Functions that operate on edges returned by the functions in the first group

Version 24.2.1 | 30.August.2024

Section 5 | Slide 13

Section 5 | Slide 14

Page 8

Skills Journal

Complete:

Section 5 Skills Journal

Lab Exercise

Complete:

Skills Journal (image designed by pngtree) Lab Exercise

Version 24.2.1 | 30.August.2024

Section 5 | Slide 15