受电弓滑板状态监测的图像目标提取  

Image Object Detection in Monitoring of Pantograph Slippers

作　　者：陈维荣[1] 冯倩[1] 张健[1] 于国旺[2] 李哲[2] 

机构地区：[1]西南交通大学电气工程学院,四川成都610031 [2]朔黄铁路有限公司,河北沧州062350

出　　处：《西南交通大学学报》2010年第1期59-64,共6页Journal of Southwest Jiaotong University

基　　金：朔黄铁路有限公司科研基金的资助

摘　　要：将数字图像处理技术应用于受电弓滑板状态监测,给出了受电弓无线图像监测系统的工作原理和组成结构,设计了图像目标的提取算法,将受电弓滑板图像的提取划分为粗提取和精提取2个步骤.该图像目标提取算法结合边缘提取、形态学图像处理、投影变换、聚类分析和Hough变换等方法,完成复杂目标图像的提取,并实现较高精度的定位.实验结果表明,所提出的算法对复杂背景、不均匀光照和噪声等影响因素不敏感,正确识别率能达到85%.Digital image process technique was used to monitoring of pantograph slippers, and the work principle and structure of a pantograph＇s wireless image monitoring system were introduced. An image object detection algorithm for pantograph slippers, composed of rough detection and exact detection, was proposed. With this algorithm, the target pantograph slipper can be detected based on edge detection, morphology, radon-transform, clustering and Hough-transform, and its position can be found exactly from a complex image. The test results show that this algorithm is hardly affected by complex background, asymmetric illumination, noises and so on, and an accuracy of 85% can be reached.

关 键 词：监测 图像处理 受电弓滑板 目标提取 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]