Algorithm of noise filtering in the problem of extracting the image contour for the system of artificial vision of robots

The image obtained by the robot's video cameras passes through several stages of transformation. At the first stage, the color image is converted to black and white, then with the selected threshold value is contrasted. It is assumed that the interference has the form of curves of minimum thickness (one pixel) or single points. The filter removes all noise. Next, a threshold filter is applied to the image obtained, which, unlike the  Sobel type masking algorithms, works much faster and more accurately. The program uses the operators of the Maple system.