摘要
提出了一种新的星载合成孔径雷达(SAR)图像中道路网的三步提取算法,即道路中心线检测、道路片段提取和道路片段连接。首先对含有斑点噪声的SAR图像进行两次滤波获得较细的道路中心线,然后对该二值图像进行Hough变换以获得道路片段,道路片段经过简单的初步连接后,再次运用遗传算法得到进一步连接。实践证明了这种方法的有效性。
This paper presents a new three-step algorithm for road network detection in a spaceborne synthetic aperture radar (SAR) image. The algorithm includes detection of road central line, extraction of road segments and grouping of road segments. First we acquired thin road central lines after two times of filtering of the SAR image with many speckles. Then we used Hough transformation to the bi-valued SAR image to obtain road segments. After simply initial grouping, the road segments were farther grouped using genetic algorithm. The algorithm proved to be effective.
出处
《雷达科学与技术》
2005年第3期156-162,176,共8页
Radar Science and Technology
基金
高校博士点专项科研基金资助项目(No.20030614001)
