基于仿射不变矩的电气化铁路绝缘子片间夹杂异物检测  

Foreign Body Detection between Insulator Pieces in Electrified Railway Based on Affine Moment Invariant

作　　者：杨红梅[1] 刘志刚[1] 韩志伟[1] 韩烨[1] 

机构地区：[1]西南交通大学电气工程学院,四川成都610031

出　　处：《铁道学报》2013年第4期30-36,共7页Journal of the China Railway Society

基　　金：国家自然科学基金(U1134205;51007074);教育部新世纪优秀人才支持计划项目(NECT-08-0825);铁道部科技研究开发计划(2011J016-B);中央高校基本科研业务费专项资金(SWJTU11CX141)

摘　　要：针对接触网非接触式图像检测中绝缘子夹杂异物自动识别定位问题,提出1种利用仿射不变矩进行绝缘子片间夹杂异物的识别定位检测方法。将预处理后的检测图像利用Hough变换线段检测方法对图像元素进行杆状物和绝缘子分类;通过仿射不变矩进行绝缘子类图元匹配,识别并定位绝缘子;对膨胀后的绝缘子图像进行边缘规则化,计算绝缘子灰度统计图,通过对统计图中条带宽度的判断,最终可判别绝缘子是否夹杂异物。实验表明:该方法对于绝缘子片间夹杂异物定位准确,自动识别率较高,可为保证电气化铁路接触网绝缘安全检测提供参考。A novel foreign body detection method between insulator pieces for catenary contactless image detection was proposed by using affine invariant moments. The proposed method aimed to overcome the difficulties in recognizing and locating the insulators and the problems in the various global images of the suspension equipment. Firstly, Hough transform line detection was adopted to classify the shafts and insulators of the global image after preprocessing. Secondly, the porcelain bottles of the insulators were recognized and located with the affine moment invariant . Thirdly, the gray statistical curves of the insulators were obtained after morphological dilation operation and edge adjustment. Comparing the band widths of the Gray statistical curves, whether any foreign body existed were able to be distinguished. The experiments show that the method can exactly recognize and locate the foreign body between the pieces of insulators. The algorithms can achieve good accuracy and high automatic recognition rates, and can offer a method to ensure effective operation of electric railways.

关 键 词：绝缘子 Hough变换线段检测 仿射不变矩 形态学 异物检测 

分 类 号：U225.43[交通运输工程—道路与铁道工程]