Hi Kim,

That is an interesting question,

I think the best way to do it  is by working on illuminance only!

In Matlab you can do this by something like this:

======Code Begin========

% read test image

im_l = imread('input.bmp');

% change color space, work on illuminance only

im_l_ycbcr = rgb2ycbcr(im_l);

im_l_y = im_l_ycbcr(:, :, 1);

im_l_cb = im_l_ycbcr(:, :, 2);

im_l_cr = im_l_ycbcr(:, :, 3);

%use im_l_y to in your KSVD code to get im_KSVD_y

%re generate RGB image

im_h_ycbcr = zeros(size(im_KSVD_y));

im_h_ycbcr(:, :, 1) = im_KSVD_y;

im_h_ycbcr(:, :, 2) = im_l_cb ;

im_h_ycbcr(:, :, 3) = im_l_cr ;

im_h = ycbcr2rgb(uint8(im_h_ycbcr));

% show the images

figure, imshow(im_h);

======Code End========

Hope this helps

Ala'