摘 要
智能交通系统(ITS)是目前世界交通运输领域正在研究和广泛关注的课题。 近年来，智能交通系统的应用给交通运输业带来了巨大的经济效益，对于道路设 计、流量监控和高速公路管理起到了越来越重要的作用。论文所研究的视频车辆 检测技术在ITS中占有很重要的地位，与传统的车辆检测方法相比，视频车辆检测技术不仅具有安装维护便捷且费用较低、可监视范围广等诸多优点，同时可对道路现场图像进行智能化分析和处理。
本文利用OpenCV中对运动物体检测的数据结构、函数库，建立了一个视频车辆分析系统，用于道路上车辆的检测。检测过程是首先对视频图像进行预处理，比如二值化、去噪等；然后进行背景的提取及更新，由于背景差分法是通过当前帧和背景帧相减来提取运动目标，所以实时的背景更新尤为重要，本文采用多帧求平均的方法来更新背景，避免了光照条件和气候环境等带来的不利影响；最后利用背景差分法检测出运动车辆。经过实验验证，该方法可以较准确地检测出车辆目标，检测的成功率可达到90%以上。
关键字： 视频处理 车辆检测 OpenCV
ABSTRACT
Intelligent Transportation Systems (ITS) is the subject of study and attention in the world's transportation sector.In recent years, the application of intelligent transportation systems has brought enormous economic benefits to the transportation industry,and it has played an increasingly important role in the road located design, traffic monitoring and highway management.The video vehicle detection technology in this thesis study plays an important role in ITS.Compared with the traditional method of vehicle detection, video vehicle detection technology has easy installation and maintenance,low cost, wide range that can be monitored and many other advantages , while it can do intelligently analysis and processing to the image of the road scene.
This thesis sets up a video vehicle analysis system for detecting vehicles on the road by the moving object detection data structure and function library in OpenCV. Detection processes include:First,video image pre-processes,such as binary,removing noise;Next,background extraction and update, as background subtraction is based on that subtract the current frame from background frame to extract moving target,real-time background update is particularly important, we use multi-frame averaging method to update the background,which avoids the the adverse impact of light conditions and climatic and environmental;Finally, the detection of moving vehicles by background subtraction.Experimental results show that this method can accurately detect the vehicle target, the success rate of detecting is up to 90%.
Key words: Video Processing Vehicle Detection OpenCV
目 录
第一章 绪论 ????????????????????????????1 1.1 课题背景和研究意义???????????????????????1 1.2 国内外研究现状?????????????????????????1 1.3 OpenCV概述?????????????????????????2 1.4 本文研究内容??????????????????????????4 1.5 本文章节概括??????????????????????????5 1.6 本章小结????????????????????????????5 第二章 基于视频的车辆检测技术概述?????????????????6 2.1 引言??????????????????????????????6 2.2 视频车辆检测技术实例发展历程??????????????????6 2.3 视频车辆检测技术未来趋势????????????????????7 2.4 本章小结????????????????????????????7 第三章 车辆检测算法设计??????????????????????9 3.1 引言??????????????????????????????9 3.2 车辆检测算法流程图???????????????????????9 3.3 图像预处理??????????????????????????10
3.3.1 设定感兴趣区域?????????????????????10 3.3.2 高斯低通滤波处理????????????????????10 3.3.3 二值化阈值分割?????????????????????10 3.4 背景初始化与更新???????????????????????11 3.5 提取前景运动目标???????????????????????11 3.6 本章小结???????????????????????????12 第四章 实验结果与分析??????????????????????13 4.1 引言?????????????????????????????13 4.2 实验配置???????????????????????????13 4.3 实验结果与分析????????????????????????13 4.4 本章小结???????????????????????????16 第五章 总结与展望????????????????????????17 5.1 本文总结???????????????????????????17 5.2 工作展望???????????????????????????17 谢辞???????????????????????????????19
参考文献?????????????????????????????20 附录??????????????????????????????21