Edge Detection filter
I’m reading about the HOG algorithm now. It is said that a filter in the form of the following matrix is used to detect the vertical edge: [[1, 0, -1], [2, 0, -2], [1, 0, -1]] ...and the flipped one for horizontal. I understand that positive and negative numbers should detect significant difference in values on the right and left, and therefore detect edges. But why are 2 and -2 in the middle instead of 1 and -1?
Log in with Google