摘 要 为推进铁路现代化建设，实现货车列检作业环节的减少，推动从人检到机检、静态检测到动态检删的转变，提高列检效率和质量，保障货物列车运行安全，铁路列检站急需要一种可以对目前运行的铁路货车车辆的标签进行识别，对车底重要部位进行高速、清晰的图像采集，井可阻实现/W .d动识别的设备，因此-货车运行故障动态图像检测系统（Troubic of Mwmg Frcigjit Car Detection Systcm，TFDS）孕育而生。 本文主要介绍了铁路货车运行故障动态图像检测系统的研究过程作者将TFDS系统分为前端控制与采集于系统和计算机视觉故障自动识别子系统两部分，每个子系统又分为若干模块分别设计。本文重点对静端控制与采集子系统的车号识别设备、图像自动采集设备，以及对计算机视觉自动识别子系统的针对两类常见故障的计算机视觉图像自动别算法进行了研究。研究车号自动识别设备时，应用了射频识别(RadioFrcquency Identification，RFID)技术中的微波反射调制解调原理。其中载波信号生成子模块产生910.1MHz、912.1MHz和914.1MHz频点的可选载波信号，标签将车号编码信息通过幅度调制加载到载波信号中，反射波经过与载波同频本振信号混频并滤波后得出车号编码信号。该信号经过解码模块解码得出车号信息，研究图像自动采集设备时，本文选#了UP.8001业摄像头+Matrox Metcor】I /Digital图像采集卡的组合。在针对铁路货车两类常见故障的计算机视觉故障自动识别算法方面，采用了计算机视觉（Computer Vision，cv）技术。主要使用了直方图的灰度变换将待检图片进行预处理然后使用Sobel算子对图像进行边缘检测。在故障识别算法的设计中运用了图像投影理论，Hough变换以及Hough逆变换经典算法。铁道部tbr3070-2002铁路机车车辆自动识别设备技术条件铁路标准为本文依据。关键诃：车号自动识别系统；RFID;计算机视觉：货车运行故障动态图像检测系统 Abstract In order to advance railroad modernization, realize the freight train inspection link reduction, promote from the people seized to machine inspection, static detection to dynamic check delete change, improve train inspection efficiency and quality, guarantee of the safe running of the train, the railway train inspection station in urgent need of a can now run a railroad freight car tag identification of important parts, car high speed, clear image acquisition, well resistance /W .d automatic identification equipment, therefore - Van operation fault dynamic image detection system ( Troubic of Mwmg Frcigjit Car Detection Systcm, TFDS ) born. This paper mainly introduces the railway freight operation fault dynamic image detection system research process author TFDS system is divided into a front control and data acquisition system and computer vision fault automatic recognition system in two parts, each subsystem is divided into several modules were designed. This paper focuses on the static end control and data acquisition subsystem number identification equipment, automatic image acquisition equipment, as well as on computer vision automatic recognition system for two kinds of common faults of computer vision image automatic recognition are studied. Study of the automatic car number identification equipment, application of radio frequency identification ( RadioFrcquency Ident ification, RFID ) technology in the microwave reflection modulation and demodulation principle. Where the carrier signal generation module to generate910.1MHz,912.1MHz and 914.1MHz frequency point optional carrier signal, the label will be coding information is loaded into the carrier signal by amplitude modulation, reflection wave with the same frequency of the local oscillator signal carrier mixing and filtering after the coding signal. The signal is decoded by decoding module of vehicle number information, research image automatic collecting device, this paper selected # UP.8001camera +Matrox Metcor I /Digital image acquisition card combinations. In two kinds of common troubles in railway freight car computer vision fault automatic recognition algorithm, using computer vision ( Computer Vision, CV ) technology. The main use of the histogram of the gray level transformation will be seized image preprocessing and then use the Sobel operator image edge detection. In a fault identification algorithm for use in the design of the image projection theory, Hough transform and Hough transform algorithm. Ministry of railway tbr3070-2002railway locomotive vehicle automatic identification equipment technical conditions standard is based on the railway.The key to: automatic train identification system; RFID; computer vision: Van operation fault dynamic image detection system目录摘 要1Abstract2目录4第1章TFDS故障自动检测系统的整体设计51.1系统功能51.2系统组成51.2.1概述61.2.2系统原理71.2.3系统功能71.3系统工作流程 ，81.4安装使用情况81.4.1安装和使用81.4.2重要作用8第2章前端控制与采集子系统研究92.1前端控制与采集子系统总体设计92.2车号自动识别设备102.3初末级放大子模块11第3章 图像采集设备123.1工业摄像头123.2图像采集卡133.3上位工控机133.4补光光源14第4章 计算机视觉故障自动识别子系统研究154.1计算机视觉故障自动识别子系统总体设计154.2心盘螺丝丢失识别算法16第5章 系统调试及性能参数175.1车号自动识别设备调试175.2图像自动采集设备调试195.3系统整体联调205.4系统性能参数235.4图片质量245.5设备问题24结束语25参考文献27第1章TFDS故障自动检测系统的整体设计1.1系统功能 本课题所研究的货车运行故障动态图像自动检测系统(TFDS)，主要负责对一些直接影响列车行车安全的关键部位进行图像采集和自动检测，以弥补传统列检作业方式的不足。本系统要实现的功能如下： (1)对行进中的铁路货车的转向架心盘、转向架静J动粱等重点待检测部位进行抓拍，自动生成顺位图像： (2)自动计轴、计辆； (3)自动测算车速； (4)采集车次、车号、车种车型； (5)对抓拍的重点检测部位照片进行故噔自动识别。1.2系统组成TFDS系统主要由检测信息采集设备、信息处理传输设备、列检所检测中心和其他复示终端构成。检测信息采集设备即轨边探测设备，主要有高速摄像装置、光源补偿装置、车轮传感器、AEI地面天线等组成，主要完成过车检测、光源补偿、图像采集任务。信息处理传输设备即探测站机房内设备，主要有图像信息采集设备、车辆信息采集设备、交换机、光纤收发器等组成，主要负责对过车信息处理并控制室外设备的正常工作，将采集到的图片进行处理，并将处理后的图像数据传输到列检所检测中心。列检所检测中心设备由服务器和检测中心室内设备组成。服务器主要完成车辆和图片信息的接收、存储和管理，检测中心主要完成图像信息的显示、故障信息的收集以及工作信息的记录。其他复示终端有段、局复示中心、部查询终端等，主要起监控、查询、统计等管理作用。1.2.1概述我们将整个TFDS检测系统分为前端控制和采集子系统以厦计算机视觉故障自动识别子系统两部分分别进行设计。整个系统的组成及功能框图如图2 1所示。其中，前端控制和采集子系统主要由车号自动识别设备和图像自动采集设备两部分组成。车号自动识别设备负责自动采集并识别固定在铁路货