Color distortion correction without noise boost using multi-resolution analysis

This paper describes the novel method related to correcting color distortion in color imaging apparatus. Color images acquired from CMOS or CCD digital sensors can suffer from color distortion, which means that images from the sensors are different from those human being sees in the color space. The one of main reasons of this color distortion is crosstalk between adjacent pixels. To correct this kind of color distortion, existing methods are generally to multiply some gain coefficients in each color channel and this multiplication can cause noise boost and loss of detail information. This paper proposes the novel method which can not only preserve color distortion correction accuracy, but also suppress noise boost and loss of detail information based on the image fusion. Specifically the color and low frequency information in luminance channel are extracted from the color transformed image. And high frequency information is from the original image before color transform. The method to extract the low and high frequency information is based on the multi-resolution analysis with edge preserving filters.