Abstract:An algorithm of obstacle detection based on partition adaptive threshold is proposed to solve the problem of UUVobstacle-avoidance sonar autonomously selecting segmentation threshold to detect obstacles. Firstly, the obstacle avoidance sonar im-age is evenly divided into image blocks with the same size, and the high and low thresholds values of obstacle image segmentation inthis area is determined based on Otsu for each image block. Then, the isolated noise points of detected obstacle are removed throughmorphological operation, and the connectivity analysis and hole treatment of the target area are carried out. Finally , the complete ob-stacle profile information is obtained. The data of lake trial has validated the effectiveness of this method for obstacle detection of sO-nar.
