A Fast Image Enhancement Algorithm Using Bright Channel

After summarizing the poor-illumination image enhancement methods and analyzing the shortcomings of the currently well-performed multi-scale Retinex algorithm, this paper proposed a new image speedy algorithm with detailed illumination component information. It combined illumination imaging model with target reflection features on RGB color channel, raised a new bright channel concept, and obtained computation method of illumination components by analysis. Then, illumination components were gained precisely through image bright channel gray-scale close computation and fast joint bilateral filtering. Consequently, target reflection components on RGB channel could be solved by illumination/reflection imaging model. The proposed algorithm can get excellent edge details through simple and quick computation. After being removed from the illuminative effects, the images gained are natural-colored, highly visible, and with no halo artifacts. This paper also resolved color casting problem. Compared with NASA method based on multi-scale Retinex, the proposed algorithm improved computation speed, received vivid colors and natural enhancement result.