How can I remove the dc component from an Image?

I wrote this code, it is right?

subplot(2,2,1);
I = rgb2gray(imread('L3S3T2.jpg'));
imshow(I);
FT = fft2(I);

FT(1,1) = 0;

subplot(2,2,3);
I2 = ifft2(FT);
imshow(I2);

Answer 1

[Tricky]

If I'm not wrong the dc component removal from image can also be done like this

I = rgb2gray(imread('L3S3T2.jpg'));
 dc=mean2(I);
  I1=I-dc;
imhsow(I1);

Can someone explain if there is any difference between the above mentioned Fourier method and this method.

[Bjorn Gustavsson]

Numerical precision and computation time - though the second might not be too bad...

...I can't see any reason to go through all that fft-ing just to remove the average.

[Tricky]

but why the output images from the both methods are different

Answer 2

[Ivan van der Kroon]

class input for fft2 should be double or single and you are using unit8. I get no errors, but this is probably not the fft2 you desire. Remember that fft2(I)=fft(fft(I).')

You are right though, that the (1,1) element is the zero-frequency or dc. But if this is the only thing you want to do just subtract the mean.

Answer 3

[Image Analyst]

Just subtract the mean gray level, being sure to cast the image to single to allow negative values:

meanGrayLevel = mean2(I); % This is a double.
no_DC_Image = double(I) - meanGrayLevel; % Casting I to double is necessary!!!

It's not necessary to use FFT to eliminate the DC components. But if you did want to do it that way, you'd just set the first element of the FT image to zero, and then inverse transform, like you did.