基于HOG特征和SVM的绝缘子识别与定位  

Insulator Location and Recognition Algorithm Based on HOG Characteristics and SVM

作　　者：李岩 

机构地区：[1]青藏铁路公司,西宁供电段,西宁810006

出　　处：《交通运输工程与信息学报》2015年第4期53-60,共8页Journal of Transportation Engineering and Information

基　　金：国家自然科学基金重点项目(U1234203);国家高技术研究发展计划(863计划)(2011AA11A102);中央专项科技创新项目(SWJTU12CX030)

摘　　要：绝缘子识别与定位是电气化铁路接触网绝缘子故障检测图像处理的重要前提。为了解决不同现场环境下的普适问题,结合绝缘子图像的特点,提出了基于梯度方向直方图(HOG)特征量提取和SVM分类器相结合的绝缘子识别与定位方法。通过对综合检测列车现场拍摄图像进行处理,将实际背景分三类分别分析并综合优化HOG DETECT及SVM参数。结果表明,在小样本容量下得到较好的识别效果,实现了实时处理,具有实际的应用价值。Insulator location and recognition is detection in detecting breakages of insulators of an important precondition for fault catenaries of an electrified railway. The method based on the histogram of oriented gradients （HOG） algorithm and SVMwas proposed combined with the features of insulator image and to be applied to various environments. The parameters of HOG detect and SVM were optimized by dividing the background into three real scenes and by processing the images on site from the train inspection car. Results show that a good and real-time recognition effect has been achieved with a few samples. This method is of good practical useness.

关 键 词：计算机视觉 梯度方向直方图 支持向量机 绝缘子识别 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]