The key of real-time obstacle avoidance for AUV is to accurately and fast identify underwater obstacle. In this paper, according to the characteristics of the multi-beam forward-looking sonar of AUV, we extracted the obstacle information from sonar image with the image filtering, fuzzy clustering and morphological methods. Finally, it can be proved that the real-time image processing method is feasible and effective by the lake trial.
Automatic Surface Inspection System for Hot Rolled Slabs by Real Time Image Processing Method