基于DM642的图像检测和跟踪系统的设计与实现

【作者】 廖军；

【导师】 管庆；

【作者基本信息】 电子科技大学 ， 信息与通信工程， 2007， 硕士

【摘要】 随着智能系统技术的不断发展和分析技术的不断进步,各种视频监控系统的研究和开发在军事和民用等诸多领域得到了广泛的关注,而运动目标的检测和跟踪技术,因其广泛的应用前景也成为一个重要的研究课题。本课题“基于DM642的图像检测和跟踪系统的设计与实现”的研究目的就是在DSP技术基础之上,实现一种嵌入式的数字视频监控系统,完成系统的软硬件设计,同时探讨研究一种实用的动态复杂背景下运动目标检测和跟踪算法。本文的主要工作和研究成果如下:1.动态复杂背景下可见光刚体目标检测和跟踪算法研究:因摄像机颤动而引起的全局运动是在图像序列中普遍存在的,这个因素在图像中将产生平移,旋转和缩放的后果,所以图像校正过程也就成了在运动分析和分割中关键而不可分隔的一部分。而全局运动可以用含有几个参数的模型来表征,如六参数、八参数模型等。本文采用六参数仿射模型下的全局光流约束来进行运动目标检测,采用能量累积的方法来进行运动目标跟踪,并完成了Matlab下的算法全功能仿真。实验结果证明了该算法的有效性和稳健性,此外该算法还具有良好的通用性。2.基于DSP的嵌入式视频监控系统的软硬件设计技术研究:介绍了一种基于TI公司TMS320DM642芯片的PCI视频监控系统的设计和实现,该系统可实时处理两路模拟摄像机数据或两路网络摄像机数据,讨论了系统软硬件的构成,分析了在TI的软件参考框架RF5架构下的操作系统任务调度,同时探讨了基于TI的IOM(I/O Mini-driver)模型的硬件驱动开发。实验结果表明该系统能实时对多路图像数据运行监控算法,并将监控结果通过PCI总线传送给主机,可以方便的完成本地存储和远程监控,为各种视频监控算法提供了有效的软硬件平台。更多还原

【Abstract】 With development of intelligent systems and analysis techniques, various video monitoring systems have gained widely attention both in military and civil field. Especially, moving object detection and tracking has been an important research subject because of its wide application prospect.The subject“The design and implement of image detection and tracking system based on DM642”focuses on the software and hardware design of an embedded digital video monitoring system based on DSP technique, and also studies a practical algorithm of moving objects detection and tracking under dynamic complex background. The major work and achievement of this paper are presented as follows:1. A visual object detection and tracking algorithm under dynamic complex background: global motion caused by camera vibration is quite common in image sequence. This factor will result in translation, rotation and scale in image. So the image registration process has become a critical and non-separable part of motion analysis and segmentation. Global motion can be modeled by a few parameters such as six parameters model and eight parameters model etc. This paper uses six parameters affine transformation model and global optical flow restraint for moving object detection and energy accumulation for tracking, also completes the simulations on the Matlab platform. The experiment results indicate that the algorithm is effective and dependable.2. A software and hardware design technique of embedded video monitoring system based on DSP: The design of PCI video monitoring system based on TI’s DM642 is introduced, which can process two–channel analog camera or two–channel IP camera video stream. The implementation of the core system modules is discussed, including task schedule based on TI’s RF5 framework and IOM mini–driver. The experiment results indicate that this system can process monitoring algorithm for multiplex video stream and transfer monitoring results to the host through PCI bus in real time. It is convenient for user to complete the functions of local-storage and remote monitoring, and provides an effective software and hardware platform for monitoring algorithm.更多还原