基于船头特征提取和轮廓定位的港口舰船检测

针对高分辨率光学遥感图像中靠岸舰船目标难以检测的问题，提出一种新的舰船自动检测方法.首先利用支持向量机（support vector machine，SVM）对变换到极坐标系下的疑似船头目标进行分类，得到候选舰船的船头位置和舰船方向；然后利用舰船周围区域内的直线信息校正舰船方向；最后结合灰度和边缘信息检测船身，用以确定舰船目标.结果表明，该方法能够准确检测靠岸舰船，具有较好的鲁棒性. 

Detection of Ships in Harbor Based on Ship Head Feature Extraction and Contour Localization

In this paper,a new method was proposed to detect inshore ships in high-resolution optical remote sensing images.Firstly,suspected ship head targets,been transformed into the polar coordinate system,were classified with support vector machine (SVM) to obtain head positions and directions of candidate ships.Secondly,the ship direction was refined by utilizing the line segments information around the ship.At last,the ship body was detected using grayscale and edge information in order to finally confirm this ship target.Experiment results show that the proposed method can accurately and robustly detect inshore ships.