image compression using FFT

Sir how can we compress image using FFT transform..RLE coding is not suitable with the FFT..what coding technique is suitable for FFT to compress the image..

Answer 1

[Walter Roberson]

RLE is a lossless compression technique. Compression with FFT is a lossy compression technique. You do the FFT, and you throw away some of the coefficients and output the rest; then for reconstruction you let the missing coefficients be 0 and do the inverse FFT.

Which coefficients you should throw away is something for you to explore.

Answer 2

[sam k]

a=imread('link.jpeg');
grayIm =rgb2gray(a);
[row col] = size(grayIm);
subplot(2, 2, 1);
imshow(grayIm);
title('original image')
A=fft2(grayIm); %2D fft
count_pic=2;
for thresh=0.1*[0.001 0.005 0.006]*max(max(abs(A)))
ind=abs(A)>thresh;
count=row*col-sum(sum(ind));
Alow=A.*ind;
per=100-count/(row*col)*100;

Blow=uint8(ifft2(Alow)); 
subplot(2,2,count_pic);
imshow(Blow);
count_pic=count_pic+1;
title([num2str(per) '% of fft basis'])

end

[Thinh]

can you explain this, please

[Sulaymon Eshkabilov]

This means what % of the highest FFT coeffcients to keep.

It can be also applied for color (RGB) images as well:

A = imread('A1.jpeg');
Afft=fft2(A);
Asort = sort(abs(Afft(:)));
counter=0;
for Keep = [.95 .1 .05 .001]
    threshold = Asort(floor((1-Keep)*length(Asort)));
    Ind = abs(Afft)>threshold;
    Atlow = Afft.*Ind;
    Alow = uint8(ifft2(Atlow));
    s = whos('Alow');
    totSize = s.bytes;
    counter=counter+1;

    figure(counter)
    imshow(Alow)
    saveas(gcf, strcat(['FFT_IMG', num2str(counter) '.jpeg']))
    s = dir(strcat(['FFT_IMG', num2str(counter) '.jpeg']));
    filesize(counter)=s.bytes
    title([num2str(Keep) '% of fft basis is kept and updated image file size is: ' num2str(s.bytes)])
end