Hardware manual

Impact Reference Guide Polygon Smoothing
3-153 Datalogic Automation Inc.
greyscale image is updated by averaging the current image with the existing model. The model tracks
the number of images that have been added and weights the current image accordingly.
For example, if the current image is the fourth one to be added to the model, the new model pixels are
calculated by giving the current image one-fourth weight and the current model three-fourths weight.
The mask image is updated by ANDing the non-edge pixels of the current image with the existing
model mask. As a result, the model mask includes a pixel only if it was included in ALL of the images
that were added. The overall result of adding images to the model is to make the model more generous
by incorporating more "do not care" regions, thus accounting for more variations in the inspected part.
3. Train New Template With Thresholding
This mode discards all previous training and creates a new model on this image. The model greyscale
image is a copy of the training region. The model mask image is created by thresholding the image and
then eroding it by the standoff distance.
IMPORTANT: Templates made with thresholding only include pixels within the threshold range. For
example, a model of text would include only the black pixels inside the characters OR only the white
pixels in the background, not both.
4. Add and Train Template With Thresholding
This mode is the same as Add and Train Template With Edge Detection, except that the model mask is
created using thresholding.
Greyscale Difference Type
Adaptive Difference: The tool finds the mean of the grey level differences between the template and the
image. It then automatically adjusts the difference image to reflect this mean. This difference type allows the
system to ignore small fluctuations in lighting, since the difference mean moves automatically.
Fixed Difference: No adjustment is made to the difference image.
See Chapter 3 for a description of the remaining VPM tools.
Polygon Smoothing
In the Flaw Detection Drawer
The Polygon Smoothing tool is similar to the image preprocessing performed by the image filtering tools
(e.g. the Average Filter and Gaussian Filter tools). This tool preprocesses polygon boundaries for boundary
analysis methods. This tool takes a blob list, converts it to a polygon list, and smooths those polygons. Since
a polygon that is derived from a blob contains only right angles, some form of smoothing is needed in order
to perform a reasonable analysis of the boundary.
Smoothing removes jagged edges caused by pixilation from image noise and blob thresholds. This tool per-
forms convolutions of the polygon vertices with a Gaussian filter to smooth boundary polygons. When a
higher smoothing level is selected, a larger filter is used in the convolutions.
Input Name What it is
Input Image Linked Image
Tool Origin The tool’s origin relative to the entire image space
Blob List These blobs are smoothed
Smoothing Level The amount of polygon smoothing (the number of convolutions applied
to the filter before it is applied to the polygon). A higher value uses a
larger filter for greater smoothing.