A robust underwater object detection method using forward-looking sonar images

Abstract

ABSTRACT Acoustic sonars have been widely used to provide underwater perception capabilities in various scientific and engineering applications. In particular, underwater object (or obstacle) detection is one of the most common applications of acoustic sonars to enhance the autonomy of underwater vehicles operated under conditions of low or zero visibility. This study proposes a robust underwater object detection method using two-dimensional forward-looking sonar (FLS) images without prior information on target objects. A successive image registration scheme is applied to the sequential object detection results to improve the robustness of the proposed method to false-positive detection errors caused by acoustic artefacts. Experimental results are shown to demonstrate the feasibility of the proposed method using FLS images obtained in an underwater environment.