一种适用于SAR图像边缘检测的序贯算法

A Sequential Algorithm of Edge Detection for SAR Image

提出了一种用于检测SAR图像上阶跃性边缘的序贯算法。由于SAR图像上存在大量的光斑噪声而使图像质量受到严重影响 ,而光斑噪声属于乘性噪声 ,不服从高斯分布 ,因此常用的图像处理方法难以取得好的效果。文中首先根据光斑噪声模型 ,自动确定图像中高反差部分 ,然后采用具有高抗噪性能的边缘检测算子 ,进行边缘提取。实验证明该算法明显提高了特征提取速度 ,同时具备良好的抗噪声能力。

The paper proposes a sequential algorithm for step edge detection in SAR image. For lots of speckle, the SAR image becomes horribly destroyed. Because the speckle is a multiple noise, and non-gauss distribution. simple image processing method is hard to get good result. In this paper, a method is proposed: firstly high contrast regions in image are determined adaptively based on speckle model, secondly an edge detection algorithm with high anti-noise ability is used to work. The experimental result demonstrates the algorithm and has shorter processing time and better anti-noise ability.