MATLAB: I want to perform SVD on cell matrix. how to apply svd on bunch of matrix? or i have to apply svd on all single matrix? but I have so many matrix of different sizes. i have used cell function for that. now for apply svd please help me.

cellmatrixmfccmultiple wavreadsvd

here is my code:
Tw = 40 ; Ts = 20; alpha = 0.97; M = 20; C = 6; L = 22; LF = 300; HF = 3700;
filelist1=dir('*.wav');
n=size(filelist1);
P=cell(n);
for k1=1:n
filename1=filelist1(k1).name;
[l1,fs1]=wavread(filename1);
l1 = l1(:,1);
P{k1} = …
mfcc( l1, fs1, Tw, Ts, alpha, @hamming, [LF HF], M, C+1, L );
%[u s v]=svd(p); %[u s v]= svd(p(1))
end

Best Answer

Use numel instead of size, because the latter replies a vector and for k1=1:size(x) might not do what you expect.

 n = numel(filelist1);
 P = cell(1, n);  % cell(n) is a cell(n, n)
 for k1 = 1:n
   ...

Here you process the files in a loop already. Calculating the SVD in addition should be easy:

 S = cell(1, n);
 V = cell(1, n);
 D = cell(1, n);
 for k1 = 1:n
   ...
   if all(isfinite(P{k1}(:)))
     [S{k1}, V{k1}, D{k1}] = svd(P{k1});
   else
     ??? What ever you want.
   end
 end

Related Solutions

MATLAB: SVDS returns empty matrix.

http://www.mathworks.com/help/matlab/ref/svds.html

"svds checks the accuracy of the computed singular vectors. If the vectors are not accurate enough, then svds returns fewer singular values than requested. To obtain the requested number of singular values, try decreasing the error tolerance in the options structure."

MATLAB: How to apply SVD to dwt subband

Two bad things I see right off the bat. You used error and sum as the names for your variables. DON'T DO THAT!!! Those are the names of very important built in functions that you will destroy if you use them.

Why do you have your own psnr function when there is a built in psnr() function? And why, in that function, do you calculate mse when there is a built in function immse()? And why do you compute the MSE (Mean Square Error) but call it SVD (singular value decomposition)?

When I ran your code with a standard demo image:

hostimage= imread('cameraman.tif');

it ran with no errors. I don't know if it did what you want, but at least it ran and finished with no errors.

Finally why did you not give us the entire error message instead of just a snippet from it, omitting the actual error itself???

You aren't putting a color image in for hostimage, are you? I'd bet that you are. Replace your function with this version:

function SVD_C()
hostimage = imread('peppers.png');
% Get the dimensions of the image.  
% numberOfColorChannels should be = 1 for a gray scale image, and 3 for an RGB color image.
[rows, columns, numberOfColorChannels] = size(hostimage);
if numberOfColorChannels > 1
  % It's not really gray scale like we expected - it's color.
  % Use weighted sum of ALL channels to create a gray scale image.
  hostimage = rgb2gray(hostimage); 
end
k = 10;
[Uh, Sh, Vh] = svd(double(hostimage));
VhT = transpose(Vh(1:columns, 1:k));
compressed_image = Uh(1:rows, 1:k) * Sh(1:k, 1:k) * VhT;
% PSNR
[PSNR_SVD, MSE_SVD] = psnr(hostimage, compressed_image)
% Save to disk.
imwrite(compressed_image, 'lena_svd.jpg');
% Display the images.
subplot(1, 2, 1);
imshow(hostimage, []);
title('Host Image', 'FontSize', 20);
subplot(1, 2, 2);
imshow(compressed_image, [])
title('Compressed Image', 'FontSize', 20);
end

Best Answer

Related Solutions

MATLAB: SVDS returns empty matrix.

MATLAB: How to apply SVD to dwt subband

Related Question