智能视频监控中运动目标检测与跟踪技术研究及实现.pdf
《智能视频监控中运动目标检测与跟踪技术研究及实现.pdf》由会员分享,可在线阅读,更多相关《智能视频监控中运动目标检测与跟踪技术研究及实现.pdf(66页珍藏版)》请在淘文阁 - 分享文档赚钱的网站上搜索。
1、中南民族大学硕士学位论文智能视频监控中运动目标检测与跟踪技术研究及实现姓名:贾茜申请学位级别:硕士专业:通信与信息系统指导教师:谢勤岚20090422中南民族大学硕士学位论文I 摘 要基于计算机视觉的智能视频分析融合了图像处理、模式识别、人工智能、自动控制及数学和计算机科学等多个学科领域的技术,具有重要的科学意义和广阔的应用前景。随着信息技术的高速发展以及安全形势的迫切需要,人们对安防设备的性能要求日益提高。利用计算机视觉技术,提高视频监控系统的自动化、智能化是未来的发展方向。运动目标检测与跟踪是视频监控中的一个重要任务,它是后续各种高级处理的基础,如模式识别、行为分析等。现实的监控环境往往是
2、错综复杂、变化无常的,探求能够从容应对复杂环境的各种变化,快速、准确而稳定的检测和跟踪目标的方法是该课题的主要研究内容。本论文以实现智能化视频监控系统这一背景为基础,重点针对智能视频监控系统中运动目标的检测和跟踪这两个关键技术进行研究,以期获得满足智能视频监控更为实用的算法,以及系统整体的闭环控制方法。论文针对运动目标检测与跟踪算法及其发展进行了全面的综述。首先,回顾了若干常用的运动目标检测算法,包括光流法、时域差分法、背景差分法,并阐述了基于高斯背景建模的前景检测方法;然后,对常用的跟踪算法作出了分类,并深入讨论了 Mean Shift 理论及相关的目标跟踪算法。针对这些算法,探讨了各方法的
3、原理,并分析、比较它们各自的优缺点和适用范围。在此基础上,论文完成了三个方面的工作:结合帧差法和 CAMShift 算法,设计了一种自动的运动目标检测与跟踪算法。首先用时间连续的三帧双差分对运动目标识别和提取,自动的选取跟踪框;再通过 CAMShift 算法计算目标的精确位置并调整跟踪窗口大小。将上述改进的目标检测与跟踪算法完成了在 DSP 上的实现,并设计了一个由 DM642-PCI 开发板和伺服机云台组成的闭环控制系统。该系统首先自动检测运动目标,再在每一帧中通过 CAMShift 算法计算目标的精确位置;最后将目标质心与视野中心的偏差信息转化成控制协议,通过串口发送,驱动云台转动来改变摄
4、像机视野,使目标始终可见。利用 CCS2.2 开发环境,在以 TMS320DM642为核心的硬件平台上实现了系统的软件算法。设计了一种高速球形摄像机 PTZ 跟踪的控制策略。球形摄像机机能在水平、垂直、光轴三个方向运动,360范围全方位高速旋转。在球机机械参数未知的情况下,通过控制球机做间歇性转动,调整球机 P/T 方向姿态角度使被跟踪对象始终可见;当视野中心对准目标后,控制球机开始进行变倍动作对图像缩放,可以观察或抓拍目标局部细节。同时,针对球机变倍控制中跟踪窗口大小自适应调整的问题,利用 SIFT 特征匹配算法设计了一种计算球机变倍率的方法,并给出智能视频监控中运动目标检测与跟踪技术研究及
5、实现II了一种在 RGB 颜色模型中基于 Mean Shift 的跟踪算法。基于 PELCO 协议,利用 VS2005 和 OpenCV 软件平台,实现了 PTZ 跟踪的整体流程。论文的主要研究集中在算法和系统控制流程设计上,而对它们结合硬件的实现为实验性的工作,搭建了系统平台。为了发挥 DSP 的强大性能,还需对嵌入式软件进一步优化,而这是个非常需要时间和工程项目经验的工作。对于球机PTZ 跟踪策略,通过实验证明能有效、实时的跟踪目标,并能够控制球机放大拍摄目标的局部细节,在犯罪取证的应用场合有一定参考价值。由于目前对于 PTZ跟踪实现方面的参考资料较少,本文策略为预演性的研究,离真正的商业
6、化用途仍有一定差距。关键词:智能视频监控;运动目标检测与跟踪;TMS320DM642;Mean Shift 算法;高速球形摄像机;PTZ 控制中南民族大学硕士学位论文IIIAbstractComputer vision-based intelligent video analysis combines many technologiesincluding image processing,pattern recognition,artificial intelligence,automaticcontrol,mathematics and computer science,etc,and ha
7、s vital scientific significanceand broad applied prospects.Along with the fast development of informationtechnology as well as the urgent need of security situation,the requirement for higher-powered security equipment is ever-increasing.It is becoming the developmentaldirection in the future that m
8、aking video surveillance system be roboticized andintellectualize through computer vision technology.Both motion target detection andtracking are important tasks in video surveillance for they are the cornerstones of thefollowing each kind of high-level processing,like pattern recognition,behavioran
9、alysis,etc.Since the realistic monitoring environment is often complicated andconstantly changing,it is the main research content that searching the method whichcan manage various kinds of change in complex environment,and detect and tracktarget rapidly,accurately and stably.Based on the background
10、of realizing intellectualized video surveillance system,this paper mainly research on two key technologies in the intelligent video surveillancesystem,namely motion target detection and tracking.The purpose is to design the morepractical algorithm and form closed-loop controlling method in whole to
11、realize theintelligent video surveillance.The paper has given a comprehensive summary of the motion target detection andtracking algorithms and their development.Several commonly used motion targetdetection algorithms are firstly reviewed,including optical flow,time-difference,background-difference,
12、then the method of foreground detection based on the Gaussbackground model is elaborated.A classification of the motion target trackingalgorithms is also made,and tracking algorithm based on the Mean Shift is thoroughlystudied.The principles,advantages/disadvantages,and the sphere of applications of
13、these algorithms have been analyzed and compared.With this understanding,the paperhas been fulfilled following three works:An automatic tracking system has been designed by the means of combiningdifferencing-frame and CAMShift.Firstly,foregroundbackground is detected throughsequential three-frame-do
14、uble-difference.The CAMShift Algorithm is then used tocalculate the exact location of the target and adjust the size of search window.智能视频监控中运动目标检测与跟踪技术研究及实现IVThe above-mentioned improved motion target detection and tracking method hasbeen implemented on DSP.A closed-loop system based on DM642-PCI b
15、oard andservo cradle head has been also designed.To begin with,the system detect movingtarget automatically,and then the CAMShift Algorithm is used to calculate the exactlocation of the target.The information of the warp between the targets center and thefocus of the cameras eyeshot is transformed i
16、nto the protocol commands and sent viathe serial port lastly,to rotate the cradle head to change cameras eyeshot formaintaining the target visible in the scene.The software algorithms have implementedon the hardware with the core of TMS320DM642 by CCS2.2.A kind of tactic of PTZ tracking controlling
17、based on high speed sphericalcamera has been designed.Spherical camera is active along three directions of pan,tileand zoom,and it is able to rotate fleetly with 360.Under the situation of the spherecameras mechanical parameter unknown,P/T directionless posture angle of thespherical camera is adjust
18、ed through controlling it to rotate intermittently to maintainthe target visible in the scene.After the center of eyeshot aiming at target accurately,spherical camera starts to zoom for observing or shooting the target with its localdetails.Furthermore,a method of compute the zooming ratio based on
19、SIFTcharacteristic matching has been designed for solving the issue of the trackingwindows needing adaptively adjusted in the process of zooming,as well as a trackingalgorithm in RGB color model based on the Mean Shift.Based on PELCO protocol,the whole flow of PTZ tracking has been implemented by VS
20、2005 and OpenCV.In this paper,the main research focus on the algorithms and the design of thesystems controlling flow,but implementing on the hardware is an experimental workfor just building the system platform.In order to exert the powerful performance of theDSP,there is a further work to optimize
21、 the embedded software,which needs time andthe engineering experiences considerably.Regarding PTZ track strategy of sphericalcamera,the experimental results show that it is able to tracking effectively,real-timely,and can control the spherical camera to zoom image for shooting the target with itspar
22、tial details.It has certain reference value in the application of crime evidencecollection.Since few references about the PTZ tracking realization are available atpresent,the strategy in the paper is only the preview research,still has certain gap tothe true commercialized use.Key words:Intelligent
23、video surveillance;Motion target detection and tracking;TMS320DM642;Mean Shift algorithm;High speed spherical camera;PTZ controlling中南民族大学学位论文原创性声明中南民族大学学位论文原创性声明本人郑重声明:所呈交的论文是本人在导师的指导下独立进行研究所取得的研究成果。除了文中特别加以标注引用的内容外,本论文不包含任何其他个人或集体已经发表或撰写的成果作品。对本文的研究做出重要贡献的个人和集体,均已在文中以明确方式标明。本人完全意识到本声明的法律后果由本人承担。作者
24、签名:日期:年 月 日学位论文版权使用授权书学位论文版权使用授权书本学位论文作者完全了解学校有关保留、使用学位论文的规定,同意学校保留并向国家有关部门或机构送交论文的复印件和电子版,允许论文被查阅和借阅。本人授权中南民族大学可以将本学位论文的全部或部分内容编入有关数据库进行检索,可以采用影印、缩印或扫描等复制手段保存和汇编本学位论文。本学位论文属于1、保密,在_年解密后适用本授权书。2、不保密。(请在以上相应方框内打“”)作者签名:日期:年 月 日导师签名:日期:年 月 日中南民族大学硕士学位论文1第 1 章 绪 论智能视频监控(IVS,Intelligent Video Surveillan
25、ce)源自计算机视觉(CV,Computer Vision)技术。计算机视觉是在图像处理的基础上发展起来的新兴学科,在计算机科学和工程、信号处理、物理学、应用数学和统计学,神经生理学和认知科学等研究方面,在制造业、检验、文档分析、医疗诊断,和军事等领域等各种智能、自主应用方面,都有非常广阔的发展前景。计算机视觉就是用各种成像系统代替人体视觉器官作为输入敏感手段,由计算机来代替人脑完成数字图像处理和解释。它是人工智能(AI,Artificial Intelligent)研究的分支之一,最终研究目标就是使计算机能像人那样通过视觉观察和理解世界,具有自主适应环境的能力。视频监控中所提到的智能视频分析
- 配套讲稿:
如PPT文件的首页显示word图标,表示该PPT已包含配套word讲稿。双击word图标可打开word文档。
- 特殊限制:
部分文档作品中含有的国旗、国徽等图片,仅作为作品整体效果示例展示,禁止商用。设计者仅对作品中独创性部分享有著作权。
- 关 键 词:
- 智能 视频 监控 运动 目标 检测 跟踪 技术研究 实现
限制150内