DCT transformation

hello i am trying to apply DCT transformation on image of size 512*512. for applying DCT i use following code

I = im2double(norm); T = dctmtx(8); dct = @(block_struct) T * block_struct.data * T'; D = blockproc(I,[8 8],dct);

and for Inverse DCT i am using the following one

mask = [1 1 1 1 0 0 0 0 1 1 1 0 0 0 0 0 1 1 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0]; B2 = blockproc( D,[8 8],@(block_struct) mask .* block_struct.data); invdct = @(block_struct) T' * block_struct.data * T; I2 = blockproc(B2,[8 8],invdct);

here in IDCT , what is actual importance of 'mask'. here matrix shown for mask is 8*8 as i am converting the image also in 8*8 block. if i like maintain the block of 32*32, than what is the procedure for writing 'mask' matrix. so my confusion is regarding importance of masking here and how to write mask matrix for 32*32. if anybody having any idea , plz help me . thanx in adv.....

Answer 1

[Robert Brown]

Couldn't find your image, so I found 'rice.png' within MATLAB

I = im2double(imread('rice.png'));

Display the rice.png image, ising imagesc to autoscale and display the image.

figure
imagesc(I)

T = dctmtx(8);
dct = @(block_struct) T * block_struct.data * T' 
D = blockproc(I,[8 8],dct);

In your code, you had mask as a long 1x64 vector, and I think you meant to have it as an 8x8 matrix. I manually put the semicolons after each 8 elements of the mask vector, to turn it into an 8x8 matrix. More elegantly, I could have used the command mask = reshape(mask,8,8) to reshape the 1x64 vector into an 8x8 matrix. Once I made this change, your code ran successfully, and computed new image I2, which I display here using imagesc instead of image, since imagesc autoscales the image before displaying it

mask = [1 1 1 1 0 0 0 0; 1 1 1 0 0 0 0 0; 1 1 0 0 0 0 0 0; 1 0 0 0 0 0 0 0;...
    0 0 0 0 0 0 0 0; 0 0 0 0 0 0 0 0; 0 0 0 0 0 0 0 0; 0 0 0 0 0 0 0 0] 
B2 = blockproc( D,[8 8],@(block_struct) mask .* block_struct.data);
invdct = @(block_struct) T' * block_struct.data * T 
I2 = blockproc(B2,[8 8],invdct);

figure
imagesc(I2)