A Correction Method for Saturated Images Based on Separation of Luminance and Chrominance

A correction method for saturation images in YC_bC_r color space is proposed to solve the problem of chrominance distortion and details lost caused by image saturation. Neighborhood's chrominances are used to correct the saturated pixel, and the unsaturated channel information of the pixel is used as a reference. The calibration point is located on the extension line of the original saturated straight, and has the shortest distance between the extension line and the line formed from weighted average of its neighborhood's chrominances, when 1-channel saturated pixels are corrected. Since the correction is performed in the YC_bC_r three-dimensional color space, the luminance of the corrected pixels can be fixed simultaneously. The chrominance of the saturated pixel is restored by the correlation of neighborhood chrominance, and the luminance of the saturated pixel is restored by computing original values of unsaturated channels, when 2-channel saturated pixels are corrected. The neighborhood recovery method is used to the chrominance of the saturated pixels, and the gradient maintenance procedure is employed to the luminance of the saturated pixels, when 3-channel saturated pixels are corrected. Experimental results show that the method is more effective in correcting chrominance and luminance of both RAW images and processed images, and maintains the image information of the unsaturated area as much as possible.