指数加权均值比(ROEWA)算子是一种较好的适用于SAR图像的多边缘检测算子，然而传统方法无法准确定位边缘，且不能计算边缘方向。改进边缘检测流程，分析上述问题产生的原因并采用传统ROEWA算子保证边缘检测恒虚警，对ROEWA加权后的滤波值进行差分运算保证非极值抑制的准确性。利用Radon变换具有较好抗干扰性和几何解析性的性质，提出基于Radon变换计算边缘方向。针对实测SAR图像的实验结果表明，该方法边缘检测及边缘方向计算性能较好。

Ratio of exponentially weighted average (ROEWA) operator is an multi-edge detector suitable for SAR images. The detector which is unable to compute the edge direction, cannot get the exact edge position. The reasons for these problems are analyzed and the flow chart of edge detection is modified in this paper. The constant false alarm rate (CFAR) character is retained by the traditional operator and the accuracy of non maxima suppression (NMS) is improved by the difference operator. Then Radon transform is carried out to get the edge direction for its excellent geometrical analysis and noise immunity properties, and by this way, the line connection is facilitated. The experiment results based on the SAR images show that the performance of the improved algorithm is robust and efficient.