A is a matrix of size 32*32.I want to subtract elements of each 4*4 block of A from their respective 4*4 means

A is a matrix of size 32*32. out = blockproc(A,[4 4],@(x)mean(x.data(:)));

Using the above code, I am able to reduce A to 8*8 which is obtained by averaging each 4*4 blocks of A. Now, I want to subtract elements of each 4*4 block of A from their respective 4*4 means. Thanks !

Answer 1

[Matt J]

out = blockproc(A,[4 4],@(x)mean(x.data(:)));
result = A - repelem( out , 4,4);

[Matt J]

Although instead of blockproc, I recommend using sepblockfun downloadable from here. For larger arrays, you will find it much faster.

A=rand(3200);

tic;
out = blockproc(A,[4 4],@(x)mean(x.data(:)));
result1 = A - repelem( out , 4,4);
toc; %Elapsed time is 9.660623 seconds.

tic;
result2 = A - repelem( sepblockfun(A,[4,4],'mean') , 4,4);
toc; %Elapsed time is 0.053036 seconds.

[bidlee devi]

Thank you Matt.

[bidlee devi]

After running the code

result1 = A - repelem( out , 4,4);

It says

"Undefined function 'repelem' for input arguments of type 'double'."

[Matt J]

Upgrade!

https://www.mathworks.com/help/matlab/ref/repelem.html#bum31rc-1_seealso

Answer 2

[Fangjun Jiang]

A=rand(32);
out = blockproc(A,[4 4],@(x)mean(x.data(:)));

temp=repmat({ones(4,1)},8,1);
temp=blkdiag(temp{:});
Result=A-temp*out*temp'
%% or run below to see the effect
temp*magic(8)*temp'

[bidlee devi]

Thank you Fangjun