You are here: Start » Program Examples » Comparing Golden Template

Comparing Golden Template

Aim

This example shows how to find defects at the end of a blade. Golden Template technique can be used to find even the smallest defect.

Input

Set of images with a single blade.

Output

Image with marked defects on it.

Hints

Best way to find defects is to locate blade using Template Matching and compare the matches with Golden Template technique.

Use a filter CropImageToRectangle to utilize template matching matches for reducing a comparison image.

In this case both filters can be used CompareGoldenTemplate_Intensity and CompareGoldenTemplate_Edges. But CompareGoldenTemplate_Intensity provides more information about defects.

Solution (Studio)

  1. Add the EnumerateImages filter to load images from the images directory.

  2. Find the position of an object add filter LocateSingleObject_Edges and prepare a template model like on the image below:

  3. Use the match found to crop a part of the image which should be compared. Add filter CropImageToRectangle and connect it with outObject.Match.

  4. Iterate through images and save in a global parameter the output image of CropImageToRectangle which contains a not damaged blade. Name this parameter Template.

  5. Add filter CompareGoldenTemplate_Intensity and connect to inImage output of CropImageToRectangle. To the inGoldenImage connect the global parameter Template.

  6. Configure the CompareGoldenTemplate_Intensity by setting inEdgeThreshold to 35.0 and inMaxDifference to 10.0.

  7. To prepare output image add the DrawRegions_SingleColor filter and use connect to it output from the CropImageToRectangle and from the outDefects

Macrofilter Main uses the Template Matching and Golden Template techniques to find defects on a blade.

Used Filters

Icon Name Description
LocateSingleObject_Edges Detection of an object whose outlines are sharp and rigid. Often one of the first filters in a program.
CropImageToRectangle Usually used for creating images of individual objects, e.g. after Template Matching.
EnumerateImages Emulates image acquisition with images stored on disk.
CompareGoldenTemplate_Intensity Finding general object defects by analyzing brightness deviations from a template image.
DrawRegions_SingleColor Apart from drawing can also be used to "clear" a part of an image.