3*3 kernal slide

I'm working on local tri directional quienry pattern (LTriDP) for 3*3 matrix it works fine but my kernal is not moving on next 3(overlapping window) matrix elements to calculate LTDP. need help how can my kernal move after calculating firts 3*3 here is the code below.

[Walter Roberson]

Goodness! Convert that long chain of if /else if /else if statements into if /elseif/ elseif, and then use the Smart Indent feature of the editor. Your indentation makes the code hard to follow.

Answer 1

[Image Analyst]

You need an outer double loop over rows and columns of the larger matrix, then inside that you need another double loop to go over the elements of your filter at a fixed location of the larger matrix. See attached code where I do a convolution "manually" with 4 for loops instead of using the conv2() function.

Answer 2

[Walter Roberson]

Why should it move? You loop i=1:3, j=1:3 and that is it.

You calculate [r ,c]=size(LTDP) but after that you never refer to r or c so at no point are you looping over the rows or columns of LTDP.

[ayesha abbassi]

i=3; j=3 is the size of the kernal which it should traverse

[Walter Roberson]

Yes but you have no code to move the kernel. You would need one or more loops around the "for i" loop, moving the starting row and column.