论文部分内容阅读
阐述了张量投票算法基本原理,提出了一种利用张量投票算法从合成孔径雷达(SAR)图像中提取道路网的方法.首先利用一个比值算子和一个相关算子的融合检测算子将道路基元检测出来;然后再利用方向可调滤波器进行SAR图像的局部方向估计;最后对方向估计结果进行张量编码,并采用张量投票算法构建道路网.基于真实SAR图像的实验结果显示,此方法对斑点噪声不敏感,不需对SAR图像进行复杂的预处理,就能快速、准确地检测SAR图像中的道路网.
The basic principle of tensor voting algorithm is expounded and a method of extracting road network from synthetic aperture radar (SAR) image by using tensor voting algorithm is proposed.First, the fusion operator of a ratio operator and a correlation operator Road primitives are detected, and then the direction-adjustable filter is used to estimate the local direction of the SAR image.Finally, the tensor coding is performed on the result of the direction estimation and the road network is constructed using the tensor voting algorithm.Experimental results based on real SAR images show This method is insensitive to speckle noise and can quickly and accurately detect the road network in SAR images without complicated preprocessing of SAR images.