Robust Object Profile Extraction Based on Multi-beam Sonar

Abstract

Multi-beam sonar is one of the most important visual perception tools for underwater medium and longdistance object detection. The sonar image obtained by it is an important data source for Three-dimensional reconstruction of underwater scenes. In order to solve the problems of background noise and clutter in underwater target detection based on multi-beam sonar, a robust object profile extraction method for multi-beam image sonar is proposed in this paper. Firstly, threshold processing and non-maximum suppression are carried out for the points on each column of the image according to the pixel gray value to obtain the candidate points of the object profile. Secondly, according to the imaging principle of multi-beam sonar, the likelihood probability model of candidate points and the jump penalty model between adjacent columns of image are established to remove the influence of noise on profile extraction. According to the characteristics of the established objective function, the dynamic programming method is used to solve the problem. The experimental results of measured and simulated data show that this algorithm is simple and easy to operate, and can accurately and effectively extract the single profile of multi-beam image sonar, which has a high application value.