CODEWITHSUNDEEP
matlab
Ted Mosby
Ted Mosby

Reputation: 1456

Split 3d matrix into multiple 2d matrices

I have a 3x1 cell array that looks like this:

x={rand(256,901,160);rand(256,901,160);rand(256,901,160)};

    [256x901x160 double]
    [256x901x160 double]
    [256x901x160 double]

I'd like to take the 3rd dimension and split it into 160 different 2d matrices, so 160 [256x901] matrices. I want to do this because i have written different functions that take in 2d matrices.

my desired output would be a 3x160 cell array containing matrices of 256x901.

I know i need to use reshape or mat2cell but I'm not too sure on the syntax.

EDIT

I found a slower way to do it but its not great..

for i = 1:length(x)
    for k = 1:160
      y{:,k}= x{i}(:,:,k);
    end
 end

any suggestions to improve speed?

Upvotes: 1

Views: 573

Answers (3)

S Geurts
S Geurts

Reputation: 118

I tried some different approaches but I found that your own implementation is the fastest even compared to the the other solutions given. If you want more speed, you can consider a parfor loop. (parallel computing)

Upvotes: 1

EBH
EBH

Reputation: 10450

Here is an example with smaller arrays, it works on your example too:

x = {rand(3,4,5);rand(3,4,5);rand(3,4,5)};
y = cell(length(x),size(x{1},3));
for k = 1:length(x)
    t = reshape(x{k},[size(x{1},1) size(x{1},2)*size(x{1},3)]);
    y(k,:) = mat2cell(t,size(x{1},1),ones(size(x{1},3),1)*size(x{1},2));
end

on my computer it takes 0.36035 sec with your cell array.

Upvotes: 1

Toothpick Anemone
Toothpick Anemone

Reputation: 4644

How's this?

x ={rand(256,901,160), rand(256,901,160), rand(256,901,160)}';

x_1 = x{1};
x_2 = x{2};
x_3 = x{3};

two_d_arrays = cell(3, 160);

for k = 1:size(two_d_arrays, 2)   
    two_d_arrays{1, k} =  x_1(:, :, k);
    two_d_arrays{2, k} =  x_2(:, :, k);
    two_d_arrays{3, k} =  x_3(:, :, k);
end

Upvotes: 0

Related Questions