Research on Underwater Target Detection Method Based on Improved MSRCP and YOLOv3

Abstract

With the increasing demand for marine resources, the development of underwater visual robot technology has become more and more important as the main tool for developing marine resources. Aiming at the problem of insufficient illumination in the underwater environment and slow detection speed in the detection of marine biological targets, a rapid marine organisms detection scheme based on the improved MSRCP image enhancement algorithm and YOLOv3 is designed. First, in order to solve the problem of fuzzy and low contrast of underwater images, an improved MSRCP image enhancement algorithm based on a self-adaptive algorithm is proposed to clarify the collected underwater degraded images. We chose YOLOv3 as our detection method,which has high processing speed and high detection accuracy. A large-scale underwater biological data set is established,and we use the Darknet-53 network to train the data set. Finally, we use the trained network to detect the enhanced image and compare it with other target detection algorithms. Experimental results show that the proposed scheme can quickly and accurately detect underwater targets.