Using 3D contours and their relations for cognitive vision and robotics