Research on underwater polarization distribution reconstruction method for partial occlusion environment

Abstract

Underwater polarization imaging technology has important application prospects in marine scientific research, seabed resource exploration, seabed topography drawing, underwater archaeology and other fields. However, underwater polarization images are subject to degradation in image quality during the acquisition process due to occlusion by water grass, fish, and the absorption of water particles in the ocean. In order to enhance the quality of polarization images and restore distorted information, In this paper, an underwater rotating polarization imaging detection system is designed, which can capture sky polarized light about 10 m in the ocean. At the same time, an image reconstruction algorithm based on intra-frame prediction is proposed. The algorithm utilizes the temporal and spatial correlations of the angle of polarization image sequence to reconstruct distorted the angle of polarization images by predicting the current frame. Specifically, the proposed algorithm uses the surrounding information to predict the angle of polarization image of the occlusion area, and optimizes it through SATD (Sum of Absolute Transformed Differences) to obtain a better prediction residual image. Finally, the prediction residual image is combined with the prediction results to obtain the final reconstructed the angle of polarization image to achieve the purpose of navigation. Simulation and experimental results show that the proposed algorithm can adapt to partially occluded visual field environment, remove random occlusion and restore image detail information. Compared with the image before repair, the image information after repair is improved by about 65 %.