角点配准与图像差分的接触网绝缘子故障检测  

Fault Detection of Catenary Insulator Based on Corner Matching and Image Differencing

作　　者：郭晓旭[1,2] 刘志刚[1] 张桂南[1] 高仕斌[1] 

机构地区：[1]西南交通大学电气工程学院,成都610031 [2]铁道第三勘探设计院集团有限公司,天津300142

出　　处：《电力系统及其自动化学报》2015年第2期8-14,共7页Proceedings of the CSU-EPSA

基　　金：国家自然科学基金重点项目(U1134205);铁道部科技研究开发计划(2011J016-B)

摘　　要：针对接触网棒式绝缘子分布及故障特点,该文基于Harris角点与图像差分实现了绝缘子的抗旋转匹配和故障检测。首先对选取的模板图像及待检测图像分别进行Harris角点检测;再利用序贯相似性检测算法SSDA(sequential similarity detection algorithm)对模板图像及待检测图像内角点匹配;接着对匹配后的角点进行模糊聚类,由于匹配会造成有用角点的丢失,故需对角点进行恢复处理;然后利用最小二乘法进行直线拟合;最终根据直线两侧图像的对称特性,差分图像并统计纵向灰度获取故障信息曲线,给出故障判断。实验结果表明:文中方法简单有效、易于实现。Aiming at the distribution and characteristics of fault of the catenary rod insulator, anti-rotation matching and fault detection of insulators based on Harris comer and image differencing are realized in this paper. Firstly, the template image selected and the image to be detected were detected in Harris comer respectively. Then comers in the template image and the image to be detected were matched by sequential similarity detection algorithm （SSDA）. Third- ly, fuzzy clustering was used in the matched comers, and the process of comers recovery was necessary because the useful comers were lost in matching process. Fourthly, least square method was used in line fitting. Finally, the fault judgment was given by difference image and getting failure information curve under counting longitudinal gray accord- ing to the symmetrical characteristic on both sides of the image. The experimental results indicate that the method put forward in this paper is simple, effective and easy to implement.

关 键 词：棒式绝缘子 故障检测 HARRIS角点检测 序贯相似性检测算法 模糊聚类 

分 类 号：U226.81[交通运输工程—道路与铁道工程]