摘要
针对高速铁路接触网支撑结构中旋转双耳耳片断裂的问题,提出一种基于尺度不变特征变换(Scale-invariant Feature Transform,SIFT)的图像检测方法。利用待分析接触网支撑与悬挂装置图像和标准旋转双耳图像之间的局部特征点匹配,实现旋转双耳的定位与提取,通过分析旋转双耳上边界曲线上各点的弯曲度,判断是否存在耳片断裂故障。实验表明:本文所提方法能在复杂的接触网支撑与悬挂装置图像中准确识别耳片断裂故障,不受拍摄距离、拍摄角度以及曝光度等因素的影响,且具有较高的检测效率。
To deal with fracture failures of ear pieces of rotary double ears of the support structure of catenary of high-speed electrified railways ,a new detection method based on scale-invariant feature transform （SIFT） was proposed . Positioning and extracting of rotary double ears were realized by matching of local feature points be-tween the catenary support image to be analyzed and the standard image of rotary double ears .By analyzing the bending degrees of points of the upper boundary curve of the rotary double ears , w hether a fracture failure hap-pened was judged .Experiments show as follows ：The proposed method can accurately identify fracture fail-ures in complex catenary images , w hile being not affected by shooting distances , shooting angles and expo-sures ;compared with artificial detection , this method can greatly enhance the efficiency of detection with high accuracy .
出处
《铁道学报》
EI
CAS
CSCD
北大核心
2014年第2期31-36,共6页
Journal of the China Railway Society
基金
国家自然科学基金(U1134205
51007074)
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)
关键词
耳片定位
误匹配消除
边界曲线
SIFT
positioning of ear pieces
SIFT
error matches eliminating
boundary curve
