Abstract
Low-light color images are limited in their application in fields such as security monitoring and autonomous driving due to issues such as dim brightness and blurry details. To improve the problem of dim brightness and uneven local lighting in low-light color images, and to enhance image quality to meet practical application needs, this study adopts multi-frame grouping denoising and adaptive gamma correction to unify the brightness range, combined with blind source separation denoising and Pearson growth curve adjustment of brightness. To address uneven light in a single frame, dual-scale adaptive gamma correction are adopted. Comparative experiments are conducted with multiple sets of low-light and light uneven images to validate the performance of the research method. The findings denoted that the average brightness of the enhanced image increased from the original 5.03-25.31 to 57.14-80.02, the information entropy increased from 3.26 to 6.07 to 7.05-8.19, and the image processing time was only 2.47-2.55 s. In images with uneven lighting, the average gradient increased from 12.74 to 16.47 to 25.32-27.12, and the visual information fidelity reached 0.89-0.93. Full size processing could be completed in 85.06 ms. The research method effectively solves the problems of brightness fluctuations and local overexposure in low-light images, providing a reference path for low-light image applications in related fields.