Hardware manual

Impact Reference Guide Pinpoint Pattern Find™
3-91 Datalogic Automation Inc.
You can use any shape for the Search ROI including polygons, circles, and exclusion regions. If no
Search ROI is defined, the entire image is searched.
2. Pass/Fail Result: The tool runs automatically and updates the panel whenever you change the Search
ROI, any settings on this panel, or a new image is acquired. The result is indicated by the Pass/Fail
field just below the image.
The Pass/Fail result of each found pattern is shown on the image with green or red edge points. If mul-
tiple patterns are found, those with a match score greater than or equal to the Minimum Match Score
are shown in green. Patterns with a match score less than the Minimum Match Score are shown in red.
The Origin is displayed only for the pattern with the highest match score. To display the pattern on a
control panel, use the tool’s Model Display Points property.
3. Minimum Score: The match score is a normalized correlation of the found pattern and the image at
the pattern’s location. The Actual value shows the match score for the pattern with the highest score.
The Minimum Score is the lowest match score possible for a pattern to pass. If there are no patterns
found with a passing score, the patterns are displayed, but the tool fails. Use the failing pattern and
Actual score to adjust the Minimum Score to get a passing result. If the Number To Find is greater than
one, then the highest score is shown in the Actual value.
4. Number To Find: This property is a counter, not a pass/fail property. By default, the tool reports just
one instance of the pattern with a match score greater than the Minimum Match Score. To report multi-
ple instances, enter the desired number in this field.
The tool searches the image for all pattern matches, then displays the patterns for all the found
matches, up to the value of this property. When multiple matches are found, they are listed in the corre-
sponding passing or failing output property. The tool passes if it finds at least one pattern with a match
score greater than the Minimum Match Score.
Increasing this value can increase the tool’s execution time.
5. Possible Scaling: The first field shown is the Possible Scaling setting from the Train ROI panel. The
second field is the actual scaling used on the pattern. You can use the Actual Scale value as feedback to
adjust the Possible Scaling settings on the Train ROI panel. You cannot adjust the Possible Scaling set-
ting on the Search ROI and Pass/Fail panel because the tool needs to be retrained after the setting is
changed.
6. Position Accuracy: Choose whether to find the pattern location to pixel level accuracy or subpixel
accuracy. The tool runs faster using pixel level accuracy.
7. Edge Detection Sensitivity: This setting is used to calculate the gradient threshold for edge finding
when the tool searches for the pattern. To save time, the threshold is calculated once for each image
using the setting on the Train ROI panel called Edge Detection Sensitivity. The tool then uses a per-
centage of that value for the edge finding threshold when searching.
A setting toward the More end causes more edges to be detected because the calculated gradient
threshold is lower.
8. Show Image Edge Points: When you check this box, all the edge points found in the Search ROI are
displayed in the image. This causes the tool to take more time, but it helps when you are configuring
edge detection settings and debugging pattern find problems.
NOTE: Because this setting affects the tool’s speed, it is only enabled while this Setup panel is dis-
played. When you leave Setup, the setting is turned off.
Algorithm Type
When the tool runs, it uses edge detection and edge matching to find candidate match locations, followed by
correlation based final placement and scoring. The model contains both edge point lists and greyscale corre-
lation models to support both stages of the search. Edge matching is the default first stage operation because
it is fast, especially for rotated objects. However, edges are inherently variable and noise-sensitive features,
so the edge matching is complemented with correlation in the final match stages to give robust final place-
ment and scoring results.