A commonly used method of color correction for digital cameras utilizes a 3×3 correction matrix with the coefficients optimized to achieve maximum color fidelity. Although this operation leads to significant improvement in color rendering, it amplifies noise at the same time. This paper presents a new method of performing color correction in DCT domain that reduces noise amplification. The paper shows that the color correction operation performed in DCT domain is more computationally efficient than the same operation that is traditionally performed in the image domain. It is also shown that gamma correction can be applied before color correction because the introduced color distortions measured in the perceptually uniform CIE Luv color space are below the human sensitivity threshold and do not affect perceptual quality of processed images. As a result, the proposed solution leads to revision of the commonly used color processing/compression chain of digital cameras reducing its computational complexity and improving image quality.