Research on improved image recovery algorithm based on Dark-Channel and multi-scale Retinex theory
Chengyu Li, Lanyong Zhang · 2022
Due to the unique optical properties of water, the camera took the images underwater often has a series of problems such as low contrast, fog blur in the images, unclear details for noise, blue-green colour casts on the background, and so on. In order to solve these problems, this paper presents an enhanced underwater image enhancement algorithm based on dark channel prior theory and multi-scale Retinex theory(Dark-Retinex algorithms). Dark-Retinex algorithms introduce a guide filter to improve the lack of depth of field in the two-dimensional perspective rate and adapt it to the underwater environment. Aiming at the problem of colour cast, based on the multi-scale Retinex theory, the brightness component of the colour space is enhanced, the ratio of each colour channel to the brightness component in the original figure is obtained, and the grey value of each colour channel is re-determined according to the balance. Using bilateral filtering removes noise’s influence on image quality and reduces the impact of noise on image detail. At the end of the paper, the Dark-Retinex algorithm and four different algorithms are simulation experiments and analyzed. The results show that the Dark-Retinex algorithm has a good effect on the enhancement of the image quality of underwater shooting, eliminates the noise and fog blur in the image, and corrects the colour deviation.