Reinhard-illumination Based Single Scale Retinex for Low Light Images Amelioration

Abstract

— Smartphones and digital cameras are essential tools for taking color images. Nevertheless, under poor lighting conditions, obtaining high-quality color images is difficult due to low brightness and uneven lighting. Various strategies have been suggested to enhance image light quality. Nevertheless, they did not produce a good result in the terms of excessive colors, halos, and loss of vital information. Therefore, the Reinhard-Illumination Based Single Scale Retinex (RISSR) algorithm is proposed in this study. The first step of the proposed algorithm involves inputting the degraded image and then predicting a single illumination map for all color channels which are altered using the Global Reinhard operator. In turn, the enhanced illumination is used to generate better color channels which are concatenated into the color image. This image is then processed with a gamma correction function and a normalization function to obtain the final improved image. The performance of the proposed algorithm was evaluated from real images under low-light conditions and judged by five metrics by comparison with ten other algorithms. Results indicated that the model presented the advantage of using the proposed algorithm with natural colors and perceived visual quality.