Thank you for your clear description of edge detection. I have a doubt, normally we get the zero-crossing point after 2nd derivation but in your lecture, we got it in 1st derivation (Derivative of Gaussian slides). Could you explain it, please?
@43:31 - The 1st derivative plot presented on the slide is not actually showing the zero-crossing of an edge in an image. It is just the first derivative of the Gaussian which also happen to look like edge zero-crossing plot. For edge detection using LOG, we need to consider 2nd order derivative of the Gaussian and then the image need to be convolved by it. Subsequently, the zero-crossings need to be identified in the output for finding out the edge locations. Hope this helps. Thanks.
Thank you for your clear description of edge detection. I have a doubt, normally we get the zero-crossing point after 2nd derivation but in your lecture, we got it in 1st derivation (Derivative of Gaussian slides). Could you explain it, please?
@43:31 - The 1st derivative plot presented on the slide is not actually showing the zero-crossing of an edge in an image. It is just the first derivative of the Gaussian which also happen to look like edge zero-crossing plot. For edge detection using LOG, we need to consider 2nd order derivative of the Gaussian and then the image need to be convolved by it. Subsequently, the zero-crossings need to be identified in the output for finding out the edge locations. Hope this helps. Thanks.