基于灰度不变性的扣件定位特征提取方法  

RAILWAY FASTENERS POSITIONING FEATURE EXTRACTION BASED ON GRAY-SCALE INVARIANCE

作　　者：侯云[1] 李柏林[1] 刘甲甲[1] 陈小艳[1] 罗建桥[1] 

机构地区：[1]西南交通大学机械工程学院,四川成都610031

出　　处：《计算机应用与软件》2015年第11期193-196,共4页Computer Applications and Software

基　　金：四川省科技计划项目(2013GZ0032);四川省科技支撑计划项目(2012GZ0102);四川省科技创新苗子工程项目(2012ZZ056)

摘　　要：针对现有扣件定位方法对不同光照条件下的铁路扣件图像适用性差、定位不准确的问题,提出一种鲁棒的、准确的铁路扣件定位检测算法。首先利用人眼视敏度特性来确定Rank变换窗口的大小,并通过改进Rank变换增强图像边缘特征;然后基于改进的LSD(Line Segment Detector)方法在增强后的图像中搜寻垫板与轨枕接触的边界线,并利用先验知识实现扣件的准确定位。实验表明,该方法定位准确且具有很好的适应性和鲁棒性。We proposed a robust and accurate algorithm for railway fasteners positioning detection to solve the problems of exisling fastener positioning method in poor adaptability o,1 railway fastener images under different illumination conditions and inaccuracy of positioning. First, it uses human visual acuity feature to determine the size of Rank transform window, and enhances image edge features by improving Rank transform. Then based on the improved LSD method it searches in enhanced image the boundaries between plates and sleepers, and achieves accurate positioning of the fastene,＇s through priori knowledge. Experimental results showed that the proposed method was accurate in positioning anti had good adaptability and robustness.

关 键 词：扣件定位 Rank变换 人眼视敏度 LSD 

分 类 号：TP3[自动化与计算机技术—计算机科学与技术]