【作者】 胡佳；

【导师】 陈志刚；

【作者基本信息】 中南大学 ， 计算机技术， 2011， 硕士

【摘要】 通过在重要场所部署视频监控设备,不仅可以实时获得安全场所的现场状态或备份取证,还具有安全可靠、节省人力资源的优点。但是纯粹的视频监控系统有明显的不足,因为监控设备获取的是运动的连续数据,其中包含大量的无用信息,典型的如交通视频监控系统。在交通视频监控系统中获得的是成千上万的车辆运行信息,如何组织与利用这些信息并从其中提取准确数据是一项艰巨的工作,如果仅仅依靠人工的方式进行提取,不仅费时费力,而且获得的数据质量不高。为此把监控技术与视频识别技术结合起来自动提取与组织关键数据是实现自动化监控,提高管理效率的重要途径。然而,视频识别技术是视频处理的难点问题之因为视频获得的是人眼可以识别的表示数据,与计算机所采用的表示方法差别很大,开发通用的计算机视觉系统是目前尚未攻克的难题。但是通过研究特定的场景进而开发以特定目标为对象的专用视频识别系统来满足特定的应用场合,已经成为当前研究的热点,这其中最为典型的是汽车牌照识别技术。本文结合汽车牌照识别技术的研究现状,将视频识别技术与GIS技术相结合,主要完成了以下工作：1)提出利用套牌车辆不能同时出现在同一地理位置的特点,通过对车辆牌照以及位置分析进行套牌车辆实时检测的思路,并以该思路为基础进行了套牌车辆检测系统的设计。2)开发了一种结合地理位置启发式信息的汽车牌照的快速识别算法。3)提出了两种能够在GIS系统中快速寻找两点之间最短路径的算法。4)以C/S与B/S相结合的方式对套牌车辆实时检测系统进行了初步实现,验证了本文设计的正确性与有效性。5)最后以基础功能测试以及完整的实验环境仿真的方式对系统进行了最终测试。更多还原

【Abstract】 The deployed Video Surveillance Equipment at important place could supply live state, backup information, more safety and also save in manpower. Data got from the Video Surveillance contains much useless information, such as Traffic Surveillance System. Our mission is to get the information of running cars from the data flow, which cost great manpower. We combine Monitoring technology and Video recognition technology to automate monitoring the traffic flow. But, video recognition technology is one of the most difficult problems in field of computer science, because of the huge difference in data form between with human and computer can identify. Currently, we can only develop some video recognition applications limited in some special conditions. The technology of vehicle license plate recognition is one of them.In our research, we propose a vehicle license plate recognition method to the specifically situation of traffics. According to recent research of vehicle license plate recognition, we combine video recognition and technology of GIS, doing the main works as bellow:1) We propose a method of detecting false vehicle license in real time, which takes the point that false vehicle license will not appear at the same location at the same time. A false vehicle license detecting system also be designed to realizes the method.2) Combining the heuristic method about geography location, we designed a fast algorithm to recognize vehicle license in video data.3) We also design two algorithms to calculate the shortest path between two points on the map.4) We combine the C/S and B/S modes to design the false license detect system, and verify the correction of our programming.5) At last, we examine the system by basic function testing and real circumstance simulation.更多还原