Error while using blockproc. Matrix dimensions must agree.

Hi,

I'm trying to use blockproc to compress a watermark that I want to embed. I'm using discrete cosine transform. I'm using blockproc for the first time, so I just copied the code off a youtube video to see how it works.

This is the code I'm using.

A=imread('C:\Users\hanuman\Documents\studies sheep\sem 7\FINAL YEAR PROJECT\project\house.jpg');
abc=imread('C:\Users\hanuman\Documents\studies sheep\sem 7\FINAL YEAR PROJECT\project\ABC watermark.jpg'); %reading the watermark
abc=imresize(abc, [96, 96]);                                                             %resizing the watermark to a number divisible by 8

f1= @(block_struct) dct2(block_struct.data);
f2= @(block_struct) idct2(block_struct.data);

J=blockproc(abc, [8 8], f1);
depth= find(abs(J)<150);
J(depth)= zeros(size(depth));
K= blockproc(J, [8 8], f2)/255;

This is the error matlab is showing:

BLOCKPROC encountered an error while evaluating the user-supplied function handle, FUN.

The cause of the error was:

Matrix dimensions must agree.

Error in dct (line 76)
  b = W .* fft(y);

Error in dct2 (line 49)
    b = dct(dct(arg1).').';

Error in Trial2>@(block_struct)dct2(block_struct.data)

Error in blockprocFunDispatcher (line 13)
        output_block = fun(block_struct);

Error in blockprocInMemory (line 80)
[ul_output fun_nargout] = blockprocFunDispatcher(fun,block_struct,...

Error in blockproc (line 243)
    result_image = blockprocInMemory(source,fun,options);

Error in Trial2 (line 8)
J=blockproc(abc, [8 8], f1);

If you know where I'm going wrong, please help.

Thank you for your time!

Answer 1

[Matt J]

Matrix dimensions must agree.

Error in dct (line 76)
  b = W .* fft(y);

It would appear that W and y are not the same size.

[sripradha iyengar]

Ummm... the only matrices I've entered are A and abc. I didn't enter W or y. And, I din't know if I can change their dimensions.

[Matt J]

Do you still get an error when you replace abc with a random 96x96 matrix? I don't.

abc=rand(96);

f1= @(block_struct) dct2(block_struct.data);

J=blockproc(abc, [8 8], f1);

[sripradha iyengar]

You're right!

I didn't convert the image to a grayscale... I took a black and white image, so in my head it was already in grayscale. Of course matlab won't consider it that way!

as soon as I converted it, the code started working fine!

thank you so much!!!!!!!!:)

[Matt J]

You're welcome, but please Accept-click the answer to signify that we've solved it.

[Matt J]

Incidentally, blockproc is often not the fastest option. You can download mat2tiles from here

https://www.mathworks.com/matlabcentral/fileexchange/35085-mat2tiles-divide-array-into-equal-sized-sub-arrays

abc=rand(4000);

tic;
J=cell2mat( cellfun(@dct2, mat2tiles(abc,[8,8]),'uni',0)   );
toc;
%Elapsed time is 12.124396 seconds.


tic;
f1= @(block_struct) dct2(block_struct.data);
J=blockproc(abc, [8 8], f1);
toc;
%Elapsed time is 15.340144 seconds.

[Matt J]

Also, this

depth= find(abs(J)<150);
J(depth)= zeros(size(depth));

is more efficiently implemented,

J(abs(J)<150)=0;

[sripradha iyengar]

ooh! Did come across that before... didn't know that it was more efficient.. Thank you for all the useful tips!:) :)

[sripradha iyengar]

I meant mat2cell, this one is new!