The image above shows a thumbnail of the interactive Java applet embedded into this page. Unfortunately, your browser is not Java-aware or Java is disabled in the browser preferences. To start the applet, please enable Java and reload this page. (You might have to restart the browser.)

Circuit Description

A very simple filter-graph with the ImageSource connected to the two famous Sobel-filters for edge-detection. These are convolution-filters with operator size 3x3 and the following matrices (check your favorite image-processing handbook for details).

SobelX (edge-detection along x-direction):

   1  2  1
   0  0  0 
  -1 -2 -1

SobelY (edge-detection along y-direction):

  -1  0  1
  -2  0  2
  -1  0  1

Note that the original Sobel filter kernels are only sensitive to gradients in one direction (low-to-high, but not high-to-low). The second filter graph shown in the applet demonstrates how to construct the 'full' edge representation of the input image. This is done by combining (here via add operations) all four possible directions. Naturally, one could use additional filters with inverted kernels, but the filter graph above does this by mirroring the input image, applying a Sobel filter, and mirroring the output image back.