I am writing at my Bachelor thesis and and need to know how the Machine Vision Sobel filter or generally all vision filter handle the pixels in the corner and at the border of the image ? They only have three or five pixels bordering it and a 3x3 filter needs nine pixels for calculating the new value.
Most algorithms either use a reflection of the pixels at the edge, or just copy the edge pixels into the border. It may depend on settings in the algorithm.
