SAR图像道路检测及其仿真  

SAR Image Road Detection And Simulation

作　　者：梅珍[1] 林伟[1] 

机构地区：[1]西北工业大学理学院数学系,陕西西安710129

出　　处：《计算机仿真》2010年第5期201-203,251,共4页Computer Simulation

摘　　要：针对图像边缘定位易发生偏移且对噪声比较敏感的问题,传统检测方法有缺点,现从小波理论出发,提出了一种基于beamlet变换的SAR图像道路线特征提取方法。首先对未经斑点噪声抑制的原始SAR图像进行非下采样小波分解,在变换域中利用beamlet变换进行道路线特征提取,最后采用识别算法检测了道路。实验结果显示方法与传统检测算子相比,具有很好的抗噪性能,检测到更多细节特征且道路识别效果更好。This paper proposed a road line extraction method based on beamlet in SAR image from wavelet, because classical method is deviate edge position, noise - sensitive and so on. The method is applied to the SAR in which the speckle noise has not been depressed,. First nonsubsampled wavelet was processed. Then in the transform domain, the road line feature was extracted based on beamlet. At last, the algorithms of recognition was given to detect the road. Compared with classical methods, it not only can caneell noise effectively, but also give more detailed features , and the recognition of roads is better. The results of the experiments demonstrated the validity of the proposed method.

关 键 词：小线变换 小波变换 线特征提取 边缘检测 识别 

分 类 号：TP951[自动化与计算机技术]