一种引入角点特征的遥感图像道路提取方法

道路是遥感图像中的一种重要目标，遥感图像自动道路提取成为一个重要的研究方向。该文在对现有道路提取算法总结分析的基础上，提出一种引入角点特征的遥感图像道路提取方法，针对城市道路网的结构特点，建立了道路模型，并对算法的结构和策略进行了描述。在提取道路直线特征和角点特征基础上得到道路段，然后根据道路之间的几何关系对独立道路段进行连接得到最终结果，有效解决了以往道路提取算法中道路交叉点断裂的问题。另外为了得到准确的道路角点信息，还对该文算法中用到的角点检测方法进行了改进。最终实验结果分析表明该文道路提取算法的有效性。

An Improved Road Extraction Algorithm Based on Corner Feature of Remote Sensing Image

Road is an important object in remote sensing image. Automated road extraction from remote sensing images is becoming an important research focus. In this paper, after a brief overview of existing methods, we propose a new road extraction algorithm based on corner feature of remote sensing image. The architecture and strategy for the method are briefly described and the details of new road extraction based on the urban road network are presented. The line feature and corner feature of the remote sensing image are used to form the road segments, then the separated road segments are finally linked to generate a road network using geometric constraints. It is robust and efficient for road network extracting from urban areas of remote sensing image. To exploit exact corner information from remote sensing image, an improved corner detected algorithm has been developed in this paper. The analysis of the final results shows that this approach has got over the original algorithm's shortcoming.