You are here: Start » Tutorial Exercises » Edge-based Template Matching: Gasket (ebtm_gasket)

Edge-based Template Matching: Gasket (ebtm_gasket)

Aim

Your task is to create a program which will be able to detect the position of a gasket.

Input

Images of a gasket with variable location and rotation.

The input images are stored in gasket_inspection directory.

Output

The location of the object.

Hints

This exercise shows a way to find an object in an image using Edge-based Template Matching. To use EbTM please use LocateSingleObject_Edges filter.

Solution (FIS)

  1. Add EnumerateImages filter to load consecutive images from a disk directory.

  2. Add LocateSingleObject_Edges filter and connect it with the input image.

  3. Open a template matching editor on inEdgeModel and select the template region as shown in the picture below.

  4. Set the Rotation Tolerance to 20.

  5. Add the outImage and outObject.Match outputs to a data preview.

  6. Perform several iterations of the program and verify if the results are correct.

  7. Set inSearchRegion as shown in the picture below to further improve the filter's performance.

  8. Add the inSearchRegion and outObject.Point outputs to the data preview to verify the search region.

  9. Add AlignRectangle filter and connect outObject.Alignment of LocateSingleObject_Edges filter with inAlignment of AlignRectangle filter.

  10. Open a rectangle drawing editor on inRectangle and select a rectangle around the gasket as shown in the picture below.

  11. Add the outAlignedRectangle output to a data preview.

Solution - Expert (FIS)

  1. Add EnumerateImages filter to load consecutive images from a disk directory.

  2. Add LocateSingleObject_Edges filter and connect it with the input image.

  3. Open a template matching editor on inEdgeModel and select the template region as shown in the picture below. Please note that you can select multiple smaller, separate regions for better rotation detection.

  4. Set the range of possible rotations from -20 to +20.

  5. In the next tab mark a reference rectangle as shown in the image below.

  6. Add the outImage and outObject.Match outputs to a data preview.

  7. Perform several iterations of the program and verify if the results are correct.

  8. Set inSearchRegion as shown in the picture below to further improve the filter's performance. Please note that the selected region must cover all possible locations of a point which falls between the template regions selected in step 3.

  9. Add inSearchRegion and outObject.Point outputs to the data preview to verify the search region.