How to divide gray scale image into 8x8 blocks and access each block separately to apply dct2 on the block

I want to divide the grayscale image which in my case is the 'cameraman.tif' into 8x8 blocks then run dct2 on each block to

later hide a binary watermark image into each block.

So far I tried blockproc and many other methods but none of them satisfy my needs.

[dharaneesh govindaraj]

I want to do same process. I dont know how to convert the image and watermark. Please help me by providing code

[Image Analyst]

@dharaneesh govindaraj, start with running and understanding my attached blockproc() demos. Then adapt them to make the function it runs fft2() instead of whatever I used. But this is essentially what Stephane does below.

Answer 1

[Stephane Dauvillier]

file = 'cameraman.tif';
im=imread(file);
imshow(im);

In order to use blockproc you have to use a function that accpts blocksstructure, that's why you need to use something like this

imModified = blockproc(im,[8 8],@(blkStruct) dct2(blkStruct.data));
imshow(imModified)

I hope this answers your question

[Ari M Salih]

But how do I access each block to add the watermarked image to it?

This works but I still need to access them

[Stephane Dauvillier]

You have a function to add your watermak ?

Let's just say you have your function has followed

function imageWithWatermak = myFunction(imageWithoutWatermark)
...
end

Then you will do

blockproc(im,[8 8],@(blkStruct) myFunction(blkStruct.data));

Answer 2

[KALYAN ACHARJYA]

image1=rgb2gray(imread('test.jpg'));
[rows colm]=size(image1);
% pading for make 8 disible, im my case rows 96 and colm 200
% 96 alread divisible by 8, no need modification: 12 
% colm 200 no need modification: 25
% Otherwise do zero pading, where required rows or colm, or both
result=dct2(image1,[12 25]);
% This size [12 25] divides the image in 8x8 blocks, each block size is 12x25
imshow(result);

There may more simpler way also.