多场景分车型交通信息视频采集系统开发研究

发布时间：2018-03-17 14:39

本文选题：多车型信息采集　切入点：视频检测　出处：《长安大学》2014年硕士论文　论文类型：学位论文

【摘要】：准确的道路交通流信息采集和符合中国国情的交通流基础理论研究已经成为我国发展智能交通系统（IntelligentTransportSystem，ITS）缓解交通压力减少交通事故以及提高道路通行能力的的关键所在随着计算机技术的飞速发展，视频分析技术越来越受到研究人员的追捧本文在分析各种交通流信息采集方法的基础上，设计了一套多场景条件下分车型交通信息视频采集系统，并在VS2010环境下配置Opencv（OpenSourceComputerVisionLibrary，开源计算机视觉库）最新版2.4.4版本开发了相应软件文章介绍了常用的几种交通信息采集方法，阐述了视频信息采集的优势和现状，对视频检测技术方面的发展现状和发展趋势进行了具体分析；详细介绍了所开发系统的操作流程，并对其进行了效果展示；重点介绍了系统开发过程中用到的关键技术：利用彩色视频提取初始背景提高了帧差法的准确度，利用基于时间的图像叠加技术更新背景来达到自适应背景更新效果，采用二值图像形态学处理来补充背景差分造成的运动车辆信息丢失，，选用直接线性变换方法实现了像素坐标与世界坐标的转换，采取少数服从多数的投票法提高车型以及车辆瞬时速度的测定准确度，采用设置参数外置的方式实现了多场景下分车型的交通信息参数检测；同时系统给出简便的系统参数设置方法和检测结果的保存方式，除此之外对实际视频进行速度车型流量检测并对结果进行了分析，指出交通流的特性和系统中可能存在的问题测试结果显示系统无论在车速车型还是流量检测方面的检测准确率均满足理论研究分析需要在此基础上对西安市西三环南二环和东二环采集到的交通视频数据结果进行理论分析，按照车型类别讨论其参数变化规律并对检测结果中蕴含的交通流原理进行了解析
[Abstract]:Accurate information collection of road traffic flow and basic theory research of traffic flow in accordance with China's national conditions have become the key to the development of Intelligent Transport system (ITSs) in China to alleviate traffic pressure and reduce traffic accidents and improve road traffic capacity. With the rapid development of computer technology, Video analysis technology is becoming more and more popular by researchers. Based on the analysis of various traffic flow information collection methods, this paper designs a multi-scene model traffic information video acquisition system. The corresponding software is developed in the latest version 2.4.4 of Open Source computer Vision Library (Open Source computer Vision Library) in VS2010 environment. This paper introduces several common methods of traffic information collection, expounds the advantages and present situation of video information collection, and analyzes the development status and development trend of video detection technology. The operating flow of the developed system is introduced in detail, and its effect is demonstrated. The key technologies used in the development of the system are emphatically introduced: the accuracy of the frame difference method is improved by using color video to extract the initial background. The time-based image superposition technique is used to update the background to achieve the adaptive background updating effect, and the binary image morphology processing is used to supplement the motion vehicle information loss caused by background difference. The direct linear transformation method is used to realize the transformation between pixel coordinates and world coordinates. A few clothes from the majority of voting method to improve the vehicle and vehicle instantaneous speed measurement accuracy, using the setting of parameters outside the way to achieve the multi-scene sub-vehicle traffic information parameters detection; At the same time, the system gives a simple method of setting the system parameters and saving the test results. In addition, the speed and vehicle flow detection of the actual video is carried out and the results are analyzed. The characteristics of the traffic flow and the possible problems in the system are pointed out. The test results show that the detection accuracy of the system in both vehicle speed and flow detection meets the needs of theoretical research and analysis. On this basis, traffic video data collected from South second Ring and East second Ring of Xi'an City are collected. The results were analyzed theoretically. According to the type of vehicle, the variation law of its parameters is discussed, and the principle of traffic flow contained in the detection results is analyzed.
【学位授予单位】：长安大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：U491.116

【参考文献】