I am trying to understand the four 1D convolution operations involved in implementation of
Laplacian of Gaussian(LoG). I have read this answer and I am also reading this pdf (See slide# 62 and 63). My current understanding is:
LoG and separate to 1D filters in x and y:
2) Take Gaussian (
g) and separate to:
3) First apply
gyy(y) to the image. That makes two 1D convolutions.
4) Then apply
g(x) to the result from
gxx(x) to the result from
g(y). That adds two more 1D convolutions, making the total four.
1) Is the above understanding correct?
2) How is this same as
3) Also, the diagram in slide #62 contradicts slide #63, in the pdf. Which one is correct?
It's actually slide#61 and slide#62.
Please see the very last comment, posted Dec 19, 2018, on the first answer.