NSCT for a non-square matrix sized 2D image?
1 view (last 30 days)
Show older comments
Hi,
I'm doing fusion of images in Nonsubsampled contourlet transform(NSCT) using NSCT toolbox.
The NSCT toolbox is available with mathworks file exchange.It includes a densct demo program which performs the decomposition and reconstruction of an image "zoneplate.png" and provides the coefficients.Its a square matrix image.
The program is working well only for a square matrix (i.e) image of same size of row and column(For example Dimension as 512x512 0r 256x256).
But if the image is not square matrix as mentioned before and its something of the type as (i.e)504X310 or 128X256 of different size of row and column I GET ERROR...
EXAMPLE: Name of my program "raz_densct.m". Below is what I get in the Matlab command window.(Matlab7.7.0 R2008b)
??? Error using ==> horzcat
CAT arguments dimensions are not consistent.
Error in ==> symext at 29
yT = [fliplr(x(:,1:ss)) x x(:,n :-1: n-p-s1+1)];
Error in ==> nsfbdec at 26
y0 = conv2(symext(x,h0,shift),h0,'valid');
Error in ==> nsctdec at 108
[xlo, xhi] = nsfbdec(x, h1, h2, i-1) ;
Error in ==> raz_decnsct at 33
coeffs1 = nsctdec( double(im1), nlevels, dfilter, pfilter );
How is it possible to solve the prob?? How to get the coeffs for a non-square matrix sized image?
Kindly reply.
0 Comments
Answers (3)
Walter Roberson
on 10 Mar 2011
Every example that I can find for Laplacian Pyramids (the first step in NSCT) shows square images. The upsampling produces a 2^N by 2^N image after N steps, which is a more restrictive condition yet.
So far, though, I have not found anything explicitly saying that square or power-of-2 images are required.
3 Comments
Walter Roberson
on 11 Mar 2011
I do not say that: I say that in the time I spent looking at the papers and diagrams, everything was square, but I did not say anything written down saying that they *must* be square. I did not understand the mathematics nearly well enough to work through to see if it could be extended to non-square matrices.
Razia
on 20 Mar 2011
1 Comment
Walter Roberson
on 20 Mar 2011
Interesting. That's a 4 x 3 aspect ratio. Perhaps other 4 x 3 aspect ratio images would work.
Subash
on 4 Mar 2014
am also using this toolbox,, but am not so gud in matlab,, may i know procedure to run this process,plzz razia.
0 Comments
See Also
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!