Fog image enhancement algorithm based on improved Retinex algorithm
Zihong Chen, Liangyan Wang, Chunling Wang, Yang Zheng · 2022 3rd International Conference on Electronic Communication and Artificial Intelligence (IWECAI) · 2022
In order to enhance the contrast, color and brightness information of the foggy image, an improved Retinex algorithm is proposed to enhance the foggy image. First, based on the Retinex algorithm, the Gaussian iterative algorithm is used to solve the total variational energy functional function, so as to effectively maintain the color constancy. Secondly, the method of combining relative gradient and absolute gradient is used to stretch the gradient of the brighter part of the foggy image. The enhanced fog image is reconstructed under the total variational Retinex theory, and the enhancement algorithm is applied to the RGB color image. Enhance the contrast and perception characteristics of the entire image, and restore the color information of the image. Use this to enhance the blurry image in foggy weather. Finally, the image is reconstructed by adding color factors to obtain a fused fog-free image. In addition, in the image preprocessing part, an image enhancement algorithm based on bilateral filtering and high-frequency lifting filtering is proposed. The simulation experiment with Matlab shows that the improved algorithm is better than the results processed when the two algorithms are used alone through the comparison of visual effects, running time and information entropy.