Non-Maximum-Suppression method- without for loops

Hi,

I've been assigned to perform the canny edge detection without using the edge function.

So far I've used the 'Gaussian' filter to smooth-en the image, the 'Sobel' filter to get the by X and by Y derivatives and got the Gradient size matrix ( asb(X)+abs(Y) )and direction matrix. Hope its good so far

What I'd like to do next is perform the non maximum suppression preferably with using the matlab matrix manipulation.

Lets say the gradient image is G the direction image is D and I want to make a thin edges image T without the need of for loops

T= all G(i,j) that are maximal in their D(i,j) direction ( if the direction is 90 degrees so if the G(i,j)>G(i,j+1),G(i,j-1) make T(i,j)=G(i,j) else T(i,j)=0 )

Would appreciate a hand ;)

Cheers, Despairy

Answer 1

[Image Analyst]

So do you want 8 T images - one for each of the 8 directions you can go from a pixel? Can't you just use circshift to shift the image, subtract it from the original, and threshold it at zero?