基于灰度和方向一致性的遥感影像道路网提取

摘要

Road network extraction from remote sensing image is a classic problem.According to the roads’ consistency of gray scale and orientation coherence of remote sensing image,this paper proposes a new method of road network’s extraction from remote sensing images.Firstly,in order to get the basic outline of the road network,this paper establishes a gray and orientation coherence criterion segmentation model by means of gray scale and directional variance information of remote sensing image.Because the road network outline exists non-road points,holes and cracks,this paper uses the methods of expansion, corrosion and others to remove clutter blocks.Finally,this paper uses the mathematical morphological operation to extract road network.The result shows,this extraction of road network method can be applied to the complex urban remote sensing image including roads,buildings,plants and other obj ects,and can obtain good effect of road extraction.%从遥感影像中提取道路网是一个经典课题，根据遥感影像中的道路具有灰度和方向一致性的特征，提出一种从遥感影像中提取道路网的新方法。首先根据遥感影像的灰度信息和方向方差信息建立灰度和方向一致性准则分割模型，由此可从遥感影像中提取基本的道路网轮廓，然后再针对道路区域存在非道路点：空洞和裂缝等情况，采用膨胀、腐蚀等操作去除杂乱物块，最后通过数学形态学操作提取出道路网。实验结果表明，该方法能够适用道路、建筑物、植被等多种复杂地物的城市遥感影像中提取道路网，且能获得较好地提取效果。