Underwater Non-uniform Illumination Image Correction Method Based on Dark Channel and Frequency Distribution Prior

Light -will gradually attenuate in the process of water medium propagation, which will lead to the problems of uneven illumination and low contrast of underwater images during underwater detection. In order to solve these problems, a new underwater non-uniform illumination image correction method is proposed in this paper. Firstly, based on the single image reflection interference separation algorithm, a new optimization algorithm is proposed to separate the reflection layer (the light reflected by the underwater object and the background) and the illumination layer (the light scattered in the water medium) in the underwater non-uniform illumination image. According to the frequency distribution of the illumination layer in the image, a low-frequency filter is introduced into the model as the constrained prior term of the illumination layer to better extract the illumination layer and accelerate its separation. At the same time, the dark channel prior is introduced to ensure that the reflector is separated more thoroughly; Secondly, due to the low contrast of the separated reflector, the adaptive gamma transform algorithm is used to enhance the reflector. Experimental results show that the proposed algorithm is effective in removing non-uniform illumination and restoring image brightness and detail information.