基于数学形态学的SAR图像道路提取

Road Extraction from SAR Image Based on Mathematic Morphology

针对低分辨率SAR图像道路目标的各种特性,利用数学形态学的方法对道路提取进行了研究.首先利用Frost算法对图像进行相干斑抑制;然后采用改进的最大类间方差选取阈值进行分割,得到一个包含道路信息的二值图像;再利用开运算和区域选取除去非道路孤立图斑;接着,利用形态学的闭运算连接断点以及对凹凸不平的道路边缘进行平滑;最后经细化和数次剪枝得到道路。实验结果表明,该方法抗噪性好,效率高,识别道路的效果良好。

In this paper, road characters in the low resolution SAR images and algorithms of mathematical morphology are applied to extract road networks from SAR images successfully . Firstly reduce the speckle noise in the image ; then the improved Method of Maximum Classes Square Error （method of ostu ） is used to get the road image binary image including road information ; secondly remove isolated small areas with morphological opening and area selecting , thirdly connect break-line and smooth the edge of road with morphological closing . Finally , determine the centerline of the road network by Morphological thinning and cropping. Experimental results show that the new Method for detecting road is very effective without the influence of noise .