CODEWITHSUNDEEP
matlab
user3051460
user3051460

Reputation: 1485

Get data in Y-Z dimension of 3D matrix

I have a 3D matrix with size is 120x256x256 where 120 is number of slices and 256x256 is size of each image slice. I want to extract the 3D matrix into 2D image (256x256). How to get the data of 3D matrix. I try to with matlab code:

Img_3D=zeros(120,256,256);
for numSlice=1:120
   Img2D=Img_3D(numSlice,:,:);
end

But it does not work. It return a matrix that size is 1x256x256, instead of 256x256 matrix. How to fix my problem. Thank you so much

Upvotes: 1

Views: 103

Answers (2)

Nras
Nras

Reputation: 4311

You can certainly solve this problem using squeeze(), it is the most easy way of removing singleton dimensions. squeeze() detects the singleton dimension(s) and removes it (them) by calling reshape() accordingly. If speed is an issue, you could speed it up by using reshape() yourself, saving some computations:

Img_3D=reshape(1:120*256*256, [120,256,256]);

t0 = tic;
for j = 1:100
    for numSlice = 120:-1:1   %// backwards for preallocation
       slices_squeeze{numSlice } = squeeze( Img_3D(numSlice,:,:) );
    end
end
t_squeeze = toc(t0);

t0 = tic;
for j = 1:100
    for numSlice = 120:-1:1   %// backwards for preallocation
       slices_reshape{numSlice } = reshape(Img_3D(numSlice,:,:),[256, 256]);
    end
end
t_reshape = toc(t0);

fprintf('squeeze-solution: %5.3f seconds\n', t_squeeze);
fprintf('reshape-solution: %5.3f seconds\n', t_reshape);

The speedup is not too much though...

squeeze-solution: 5.654 seconds
reshape-solution: 4.790 seconds

Best, Nras.

Upvotes: 1

Robert Seifert
Robert Seifert

Reputation: 25232

Just squeeze your resulting Matrix to get what you want.

Img_3D=zeros(120,256,256);

for numSlice = 120:-1:1   %// backwards for preallocation
   slices{numSlice } = squeeze( Img_3D(numSlice,:,:) );
end

basically the same can be achieved using permute as you suggested

for numSlice = 120:-1:1   %// backwards for preallocation
   slices{numSlice } = permute( Img_3D(numSlice,:,:),[2 3 1] );
end

performance-wise equal.

And after some timings it doesn't seem to matter if you permute in advance or not:

Img_3D = permute(Img_3D,[2 3 1]);
for numSlice = 120:-1:1   %// backwards for preallocation
   slices{numSlice } = Img_3D(numSlice,:,:);
end

Upvotes: 2

Related Questions