高分辨率遥感图像港内舰船的自动检测方法

A Method for Automatic Detection of Ships in Harbor Area in High-Resolution Remote Sensing Image

在高分辨率的光学遥感图像中,港口内停靠舰船的灰度、纹理特征与陆地接近,因此同行驶于海上的舰船相比,是较难实现自动检测的目标。考虑到港口几何布局稳定的特点,提出了一套基于地理信息的舰船目标检测算法。该算法结合矢量数据和栅格数据的优点,采用模板形式存储地理信息,并以自动阈值法进行目标的粗分割,充分利用先验信息和相关知识完成并联目标的切割和断裂目标的连接。通过仿真实验,证明该算法能准确、快速的实现港内舰船的自动检测。

In high- resolution optical remote sensing images the gray- scale and texture features, of the ships docking in harbor area are similar to that of the coast, so it is difficult to automatically detect them compared with the ships at sea. In view of the stability of the geometric layout of harbor, an automatic ship detection method based on geographic information is presented. This method stores geographic information in the form of template, which integrates both advantages of vector data and grid data, then roughly segments targets using automatic threshold. After that, based on a priori information and relative knowledge, connected targets are divided and broken parts are linked. Simulation experiments show that this method can detect ship targets in harbor area accurately and efficiently.