提出一种基于局域Radon变化的SAR图像尾迹提取算法，该算法首先利用滑窗对图像进行标准差归一化处理，以突出图像中灰度与海面背景有明显差别的区域；选出突出的像素点，并根据策略将其灰度值用某些常数代替；接着在局域Radon变换中，积分值对应于这些区域线性特征的长度，通过设定阈值去除线性区域较短的非尾迹区域产生的虚警，用形态学的膨胀算法将结果中靠的很近的两条亮线合成一条，再用合理的腐蚀处理，剔除噪声信号。实测数据处理结果表明，该算法能准确提取舰船尾迹，降低虚警，能确定尾迹的起点和终点，准确计算出尾迹的边之间的夹角，而且运算简单，满足人们对系统实时性的要求。

This paper presents a method for detecting ship wakes in SAR images based on the localization of Radon transform. Firstly, the image is normalized, and the noticeable pixel would be regarded as a constant. In localized radon space a big number will correspond to a long line segment. An appropriate threshold can decrease false alarm. By choosing appropriate structuring elements, i.e. dilation and erosion lengths, the two responses which are close enough in the vertical direction will be joined into one. The effect of erosion is that responses which are too narrow in the vertical direction will disappear entirely while responses which are wide enough will be contracted vertically.The simulation and real data processing results show that the algorithm is successful in wake features detection, wake location and computational efficiency.