Research on Image Defogging Algorithm Combining Homomorphic Filtering and Retinex
Wanni Liu, Jinghan Zhang · 2024
Aiming at the disadvantages of the traditional Retinex algorithm, such as poor adaptive ability and easy to produce halos when processing images with uneven illumination, an improved homomorphic filter and multi-scale Retinex algorithm is proposed in this paper. Firstly, Sigmoid function is used to construct a single parameter homomorphic filter function, which can ensure the filtering effect and reduce the adjustment parameters. The region with large brightness difference in the image is filtered to different degrees. After the improved homomorphic filter correction, the image is enhanced by multi-scale Retinex algorithm. The experimental results show that the improved algorithm can effectively correct the illumination of foggy image, keep the original information of image without loss, enhance the contrast of low-illumination image, and make the processed image more visible.