基于多尺度形态梯度的灰度图像边缘检测  

Edge Detection of Gray Image Based on Multi-scale Morphological Gradient

作　　者：陶晓勋[1] 安如[2] 周绍光[1] 张琴[3] 李伟[3] 

机构地区：[1]河海大学土木工程学院测绘工程系,江苏南京210098 [2]中国科学院遥感应用研究所北京师范大学遥感科学国家重点实验室,北京100101 [3]河海大学水资源环境学院地理信息科学系,江苏南京210098

出　　处：《遥感信息》2007年第1期58-62,共5页Remote Sensing Information

基　　金：中国科学院遥感应用研究所;北京师范大学遥感科学国家重点实验室开放基金

摘　　要：研究了一种基于多尺度抗噪型形态梯度的灰度图像边缘检测算法。该算法通过构造5个不同的结构元素,结合多尺度合成对图像进行边缘检测。为验证本文算法的效果,用几种传统的微分算子对实验图像进行了边缘提取的实验,并把结果与用本文算法提取的结果进行了比较。结果表明:本文算法比较成功地完成了图像的边缘检测,且检测效果在完整性和细节的丰富性方面优于经典的Sobel算子,LOG算子和Canny算子,但在某些边缘的连接性和平滑性方面不如Canny算子,在这方面还有待改进。In this paper, an edge detection method of gray image, based on muhi-scle antinoise morphological gradient atgorithm is presented. In this method, 5 structure dements are built and multi-scale is applied. In order to test its effect, the results of this method and several differential coefficient operators are compared. The experiments show that the image edge is detected successfully by this method, and the effect of integrality and richness of details is better than classical operators, such as Sobel operator, Laplace and Gauss operator and Canny operator, but the connection level and smoothness of some edges are less good than the Canny operator,so this method is needed to be improved furtber.

关 键 词：数学形态学 多尺度 形态梯度 二值化 边缘检测 

分 类 号：P237.3[天文地球—摄影测量与遥感]