Ship Detection of Remote Sensing Image on FRHT and Multi-Points Curvature Based Polygon Approximation

One of the most difficult tasks of ship detection is detecting the ship which is docking at the port in remote sensing image. Traditional methods of automatic detection cannot be used to detect the land/waterboundaries, because both the gray values and textural features of a port are similar to those of the ships which are docking at the port. Therefore, ships cannot be accurately detected in this case. In this study, a novel method of land/water-boundaries detection is proposed, which is based on a polygon approximation method by incorporating two techniques, i.e., Fuzzy Randomized Hough Transform (FRHT) and Multi-Points Curvature (MPC). The method considered the feature of human vision that the straight-line of the land/water-boundaries can be detected more accurately and rapidly. With the detection result of land/water-boundaries, ships docking at the port can be accurately detected. The experiment results demonstrate that this method can achieve good result of ship detection.