基于视频图像的车辆检测和匹配跟踪方法研究

发布时间：2018-04-05 00:38

本文选题：智能交通　切入点：车辆检测　出处：《长安大学》2014年硕士论文

【摘要】：车辆的检测和跟踪是智能交通系统中必不可少的重要组成部分，它为后续的交通控制管理提供重要的数据依据。基于视频的车辆检测和跟踪较传统的检测方法，具有直观性好、抗干扰强、检测范围广和性价比高等显著优点。随着信息技术的快速发展，基于视频的车辆检测和跟踪方法的研究吸引了众多研究者的关注，成为当今的研究热点。论文介绍了视频车辆检测和跟踪技术的发展现状，给出了整个检测系统的结构框架，并结合图像处理的相关知识对视频序列进行处理，对系统中的关键技术进行了研究，具体研究内容如下：在车辆检测方面，研究了常用的背景提取方法（如，统计法、均值法和中值法）和车辆提取方法（如，光流法、帧间差分法和背景差分法）。因为利用以上方法提取到的车辆含有大量阴影信息，故又对车辆阴影的去除做了大量试验研究，最后采用一种基于阴影特征的方法去除阴影，大体步骤如下：首先确定阴影所在的方向。其中，，针对深色车辆的特殊情况，提出一种通过对比差分图像各边界灰度值方差的方法，来确定阴影的方向；然后计算出车辆阴影灰度值的分布区间；最后根据分布区间，来达到去除阴影的目的。去除阴影后，利用形态学、连通区域和图像归并的相关知识和操作，最后确定出运动目标车辆的轮廓。在车辆跟踪方面，通过研究了几种常见的跟踪方法，最后采用了一种基于区域灰度值的方法来进行匹配跟踪，并对这种方法进行了改善，利用提取到的车辆区域来确定匹配区域，从而缩小了匹配范围。结果显示，基于区域灰度值的车辆匹配跟踪效果良好。
[Abstract]:Vehicle detection and tracking is an essential part of intelligent transportation system, which provides important data basis for the subsequent traffic control management.Vehicle detection and tracking based on video has many advantages, such as good visualization, strong anti-jamming, wide detection range and high performance-to-price ratio.With the rapid development of information technology, the research of vehicle detection and tracking based on video has attracted the attention of many researchers.This paper introduces the development of video vehicle detection and tracking technology, gives the framework of the whole detection system, processes the video sequence with the relevant knowledge of image processing, and studies the key technologies of the system.The specific contents of the study are as follows:In vehicle detection, common background extraction methods (such as statistical method, mean method and median method) and vehicle extraction methods (such as optical flow method, inter-frame difference method and background difference method) are studied.Because the vehicles extracted by the above methods contain a lot of shadow information, a lot of experiments have been done on the shadow removal of vehicles. Finally, a shadow removal method based on shadow features is adopted.The general steps are as follows: first, determine the direction of the shadow.According to the special situation of dark vehicle, a method is proposed to determine the direction of shadow by comparing the variance of gray values of each boundary of the difference image. Then, the distribution interval of gray value of vehicle shadow is calculated. Finally, according to the distribution interval,To remove shadows.After the shading is removed, the vehicle contour of the moving target is determined by using the relevant knowledge and operation of morphology, connected region and image merging.In the aspect of vehicle tracking, several common tracking methods are studied. Finally, a method based on region gray value is used to match and track, and this method is improved.By using the extracted vehicle area to determine the matching area, the matching range is reduced.The results show that the vehicle matching and tracking effect based on region gray value is good.
【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】：U495

【参考文献】