Low-light color image enhancement based on Retinex

In view of the problems of low image brightness, obvious noise, poor contrast, and difficulty in obtaining detailed information in dark areas under low light environment, we propose to combine the improved particle swarm optimization algorithm with a single-scale Retinex algorithm. We convert the original RGB image to the HSI color space,and each pixel of the low light image is classified separately. The adjacent pixels are calculated with the same kernel function value. The pixels with different H values are use different filter templates to complete the image enhancement. And solve the problem of image halo effect and color distortion caused by the real-time operation of the Retinex algorithm spatial filtering. Experimental results show that the proposed algorithm performs well in brightness, contrast, and color restoration.