Luminance weighted color constancy

Color constancy is widely used in the field of image processing. It can restore the color of the object itself from color deviation image. At present, color constancy methods based on statistics mainly focus on part of the image information such as brightness information. In order to better reflect the illumination from the aspect of brightness, this paper proposes a novel color constancy algorithm based on luminance weighting, and discusses the influence of different luminance block sizes on light source prediction. The experiments demonstrate that the proposed color constancy algorithm decrease by 10.17%, 4.79% and 7.66% in mean, median and trimean angular errors compared with the optimal algorithm.