4 Non-maximum suppression Our next step is to apply non-maximum supression to the gradient information. Using the gradient information, compute the magnitude and angle of the gradient for each pixel. Then, for all pixels, compare the magnitude of the gradient of this pixel to two nearest neighbors, in the direction to and opposite the gradient direction, as defined in the table below. If the magnitude of its gradient is the largest of those neighbors, flag it to continne through the pipeline. A few notes about doing this in Matlab: - Use the atan2 function to get the angles. This returns the four-quadrant atan. - Since atan2 return values in the range of {,], shift the values to be in the range of [0,2]. by adding to them. * These may be counter-intuitive. But since the y-axis is inverted for the image coordinate plane, the angles are reflected about the x-axis. An example can be found in Figure 3 . Figure 3: Gradient Magnitude after Non-Maximum Suppression