基于LBP-HOG特征的接触网旋转双耳区域识别  

Detection of catenary rotation binaural based on LBP-HOG feature

作　　者：徐丹[1] 于龙[1] 陈唐龙[1] 王佳祺[1] XU Dan;YU Long;CHEN Tanglong;WANG Jiaqi(School of Electrical Engineering, Southwest Jiaotong University, Chengdu 610031, China)

机构地区：[1]西南交通大学电气工程学院,四川成都610031

出　　处：《铁道科学与工程学报》2017年第2期370-378,共9页Journal of Railway Science and Engineering

基　　金：四川省科技计划项目(2015GZ0148)

摘　　要：研究基于LBP特征、HOG特征与支持向量机相结合的目标定位原理,并对旋转双耳区域进行定位研究,分别使用LBP特征与支持向量机的方法和HOG特征与支持向量机的方法对旋转双耳区域进行定位,在此基础上,提出将LBP特征和HOG特征融合成LBP-HOG特征对旋转双耳区域进行定位。检测结果表明,同等条件下,LBP-HOG特征的检测效果优于单一使用LBP特征和HOG特征。In this paper, the principle of target location based on the combination of LBP feature, HOG feature and support vector machine is studied. And the positioning of the rotation binaural is investigated. The LBP feature and support vector machine method or HOG feature and support vector machine method are used to locate the rotation binaural. On this basis, the combination of LBP features and HOG features is proposed to form the LBP-HOG feature which is used to locate the position of the rotation binaural. Test results show that LBP-HOG feature effect is better than that of the single use of LBP features or HOG features under the same conditions.

关 键 词：接触网检测 LBP HOG 支持向量机 旋转双耳 

分 类 号：U216.3[交通运输工程—道路与铁道工程]