数学形态学路面裂缝识别算法研究  

Algorithm Research for Pavement Crack Recognition Based on Mathematic Morphology

作　　者：梁世庆[1] 孙波成[2,3] 邱延峻[3] 

机构地区：[1]襄樊学院,湖北襄樊441053 [2]西南交通大学峨眉校区,四川峨眉山614202 [3]西南交通大学土木工程学院,成都610031

出　　处：《路基工程》2010年第1期44-46,共3页Subgrade Engineering

基　　金：交通部西南交通建设科技项目(200431800054)

摘　　要：裂缝的自动识别是公路管理和维护系统的一个重要组成部分。根据路面图像的特点,提出了一个新的基于多结构多尺度数学形态学路面裂缝自动识别算法。首先使用不断增大的结构元素对图像进行交替开闭滤波,平滑图像并去除噪音,然后使用多尺度形态学边缘检测器提取裂缝边缘,最后使用基于标记符的分水岭变换分割出裂缝对象。结果表明,和传统方法相比,文中提出的算法显示出更好的效果。Automatic recognition of cracks is an important part of highway management and maintenance system. According to the characteristics of pavement image, this paper presents a new algorithm for automatic recognition of pavement cracks based on multi-structural and multi-scale mathematic morphology. First filter wave of the image in alternative open and close mode with the increasingly magnified structural elements to smooth image and eliminate noise; then extract crack edge with multi-scale morphology based edge detector; finally segment crack object based on watershed transformation upon tag. The result indicates that the algorithm said in the paper shows better effect compared to that of traditional methods.

关 键 词：裂缝识别 数学形态学 分水岭分割 

分 类 号：U416.2[交通运输工程—道路与铁道工程]