轨道扣件检测特征提取算法研究  

Rail Fastener Detection Feature Extraction Algorithm Research

作　　者：吴禄慎[1] 万超[1] 张丛[1] WU Lu-shen,WAN Chao,ZHANG Cong(School of Mechanical ＆ Electrical Engineering,Nanchang University,Jiangxi Nanchang 330031, Chin)

机构地区：[1]南昌大学机电工程学院,江西南昌330031

出　　处：《机械设计与制造》2018年第8期5-7,共3页Machinery Design & Manufacture

基　　金：国家自然科学基金项目(51365037)

摘　　要：对于有效的提取轨道扣件的特征,提出一种基于小波变换与多尺度局部二进制相结合的特征提取方法。首先将图像进行小波变换,可以有效的去除图像中的噪声,提取出其低频部分,突出图像局部的信息,然后使用多尺度局部二进制对其进行特征提取,而多尺度局部二进制算法则可以在既保留图像局部信息的同时,最大程度留住图像整体信息。通过实际的实验,用最小距离分类器对该特征进行分类识别,表明该方法有效,可行,提高了图像中轨道扣件的识别率。To extract the characteristics of the rail fastener more effectively,a new feature extraction method based on wavelet transform and multi-scale local binary system is proposed.First,using the wavelet transform to image,it can remove the noise effectively and extract the low frequency part in the image,and then prominent local image information, then using multi-scale local binary to get feature of the image,the multi-scale local binary count rule in retains Fig like local information at the same time,to the greatest extent retain image information. Through the experiment, using the minimum distance classifier to classify the feature, it shows that the method is effective and feasible, and it can improve the recognition rate of track fastener in the image.

关 键 词：特征提取 小波变换 局部二进制 多尺度 最小距离分类器 

分 类 号：TH16[机械工程—机械制造及自动化] TP391[自动化与计算机技术—计算机应用技术]