铁路区间通过信号机编号识别系统的研究与实现

【作者】 莫利萍；

【导师】 张修如；

【作者基本信息】 中南大学 ， 计算机技术， 2007， 硕士

【摘要】 铁路区间通过信号机编号是保证铁路行车安全的重要标识,也是定期检测中必检项目内容。研究并实现铁路区间通过信号机编号识别系统,可以极大的减轻测试工作强度,也为测试提供更完整、更可靠、更全面的数据。铁路区间通过信号机编号识别综合了图像处理和模式识别技术。论文研究了图像拍摄、图像预处理、铁路区间通过信号机柱的定位与分割、编号区域的定位与分割、编号的分割和编号识别等关键技术。在铁路区间通过信号机柱的定位、分割模块中采用了Hough快速检测边缘处理后的二值图像中的柱状物体,利用先验知识和投影直方图切割出区间通过信号机柱,并在分割出来的信号机柱图像中定位并分割出编号区域。在编号识别模块采用了模式识别中最基本的模板匹配算法,实验证明,该方法对信号机的编号识别是很有效的。本系统在现有的电务试验车厢外面安装高速照相设备,到指定距离后,触动照相设备,将拍摄的铁路区间通过信号机的图像传入识别系统进行识别。论文对铁路区间通过信号机编号识别系统中的关键技术进行了研究、提出了有效的解决策略,并给出了实例。论文提到的步骤和算法均己在软件平台上正确实现,以在电务试验车上采集的图像作为测试数据源进行测试,编号识别率达90%,满足了应用的基本要求。更多还原

【Abstract】 Railway section signal serial number is an important ID of assuring railway traffic safety, as well as the project content detected indispensably during periodical detecting. The research and implement of a recognition system of railway section signal serial number will lighten a world of intension of the test work , also provide us more correctness and completeness and comprehensive data.Both image processing and pattern recognition are adopted in the recognition of railway section signal serial number.The paper investigates some key techniques as follows: image shoot,image pretreatment,the localization and division of railway section overpassed signal post, the localization and division of serial number section,the division and recognition of serial number,etc. In the localization and segmentation module of the railway section signal post, Hough transform is used to detect quickly the pillarlike objects in the binary image after edge-handling, and transcendent knowledge and the projection histogram are used to get signal post image from which serial number districts are localized and divided. The most basically template matching algorithm of pattern recognition is adopted in the module of serial number recognition in the thesis, the experiments show that the method above-mentioned is very effective to the serial number recognition of signal.In the system, outside the existing railway S &C test vehicle is equipped with high-speed cameras, when trains arrive for an appointed distance from it, the cameras is triggered to screen the signal, and the image is input into system for recognition. The paper researched and gave an effective solution to the key techniques of recognition system of railway section signal serial number, and instances are given.Approaches and algorithms mentioned in the paper are all realized correctly in a software development platform, the experimental results used images gathered from the railway S &C test vehicle indicate the method achieves excellent performance in terms of recognition rates which is 90%,and it has met with the basical demand.更多还原