基于数学形态学的雷达图像边缘检测算法

摘要

Aiming at the problems of noises and pseudo-edges in the extraction of radar image edge, two improvements are made to the traditional mathematical morphology edge detection algorithm:first,the radar image is filtered by using nonlinear composite mathematical morphology filtering method;second,according to the characters that different elements extracted re-sult different operations,to construct a new mathematical morphology edge detection algorithm by using the idea of multi-structure elements. The experimental results show that the two algorithms solve the contradiction between signal-to-noise ratio and single-edge response to a certain extent,and realize the three indexes of Canny's edge detection performance,and keep more characteristic edge information in radar image.%针对雷达图像边缘提取中存在噪声和伪边缘的问题,将传统数学形态学边缘检测算法进行了两点改进:一是利用非线性复合型数学形态学滤波方法对雷达图像进行滤波处理;二是根据数学形态学运算中结构元素提取不同运算结果不同的特性,采用多结构元素的思想,构造一种新型的数学形态学边缘检测算法.实验结果表明两种算法在一定程度上解决了信噪比和单边缘响应的矛盾问题,实现了Canny评价边缘检测性能的三个指标,将雷达图像中更多的特征边缘信息进行了保留.