Question

I have a 3D volume and I want to apply a gaussian filter (with conv) on the first dimension.
That means that I want each column to be convolved with the filter independently.
I need something like:

result = conv(cube, Gfilter, 1 ,'same')

What I tried is padding with zeros the cube’s columns (to the size of the filter width) and then used:

conv(cube(:), Gfilter)

but this is too memory/time consuming !

Thanks
Matlabit

Answer 1

For convolution among one dimension, you can still use convn. I assume here that Gfilter is a k-by-1 array.

result = convn(cube,Gfilter,'same');