Side scan sonar image segmentation and feature extraction

Abstract

This paper describes an algorithm to make the treatment, segmentation, skeleton and characteristics extraction from acoustic images obtained from a side scan sonar. The fundamental goal is to implement a system that endows a autonomous vehicle with the capacity to know its own distance to the marine bottom and to features located on the marine environment. This features extraction would improve vehicle navigation and allow it to navigate relative to features like an underwater piper on the sea floor or a vertical wall. This paper was made based on Imagenex Sport Scan (side scan sonar) whose function is the observation of environment. Also the autonomous surface vehicle (ASV) ZARCO was used to transport the side scan sonar. Both the vehicles belong to The OceansSys Group DEEC-FEUP. A communication interface between the ASV ZARCO, Imagenex Sport Scan, and a static laptop that allows the observation of sonar data in real time is also described in this paper. The algorithms and routines implemented were validated with real acoustic images acquired during a mission. Results from algorithms application and features extraction are shown in this paper.