Ship detection at the dock using a polygon approximation method

Abstract

Ship detection at the dock is a difficult task in high-resolution optics remote sensing image processing. The gray values and textural features of the dock are similar to that of the ships. Therefore, it is difficult to extract the land/water-boundaries for the traditional methods, and it is difficult to obtain accurate detection results. In this paper, we propose a novel method of land/water-boundaries detection. This method is based on a polygon approximation algorithm which is combined by two parts, i.e., Fuzzy Random Hough Transform (FRHT) and Multi-points Curvature (MPC). The main straight-line of the land/water-boundaries can be detected accurately by FRHT and MPC, and the port polygon will be approximated accurately by the proposed polygon approximation algorithm. Experimental results demonstrate the effectiveness and feasibility of the proposed method for ship detection at the dock.