Step One - Edge Detection

The program uses a simplified Canny edge detector to locate the edges in a single frame. The edge detector is "simplified" in that it does not compute the exact location of the maximum gradient, but rather it uses the gradient as a distribution range for the location of the objects. This has two nice properties. First, the algorithm yields thicker lines which allow the tracking to be smoother. Secondly, the code is significantly faster since it does not have to calculate the maximum gradient every time.

Original Frame:

After edge detection:

Now, let's see what the color segmentation does in the next step.