【作者】 刘向；

【导师】 陈星；

【作者基本信息】 大连理工大学 ， 模式识别与智能系统， 2010， 硕士

【摘要】 随着科学技术的发展与公共安全保障需求的提高,智能视频监控系统在工业生产、日常生活、警备与军事方面的应用越来越广泛。迫切的军用和民用需求,推动着智能视频监控技术持续而迅猛的发展。视频监控的常规方案是通过专用采集卡实现图像采集再由计算机处理应用的方法,但是由于成本、体积等多方面的约束难以得到广泛的实际研究与应用,因而设计一种轻便小巧的视频监控系统成为市场所需。本文分析了智能视频监控系统的现状,比较FPGA、DSP和ARM三种芯片作为图像采集处理系统核心芯片的优缺点,结合项目背景和应用需求,提出了基于SOPC的视频图像输入与移动目标检测的实现方案,并对整个方案进行了深入研究。在该方案中各个外围芯片的配置和控制都由FPGA的逻辑实现,而对于移动目标的监控算法处理交给由NiosⅡ完成。为了满足FPGA中逻辑与NiosⅡ之间的通讯,编写了可以将NiosⅡ和控制逻辑无缝连接的符合Avalon总线规范的IP核,从而构造一个片上可编程系统。另外,在智能视频监控系统中,移动目标的检测是基础保证,本文在研究了几种常用的运动目标检测算法,并分析它们各自的优缺点,选取了易于实现的减背景法。再根据减背景法的特点,研究了背景提取和背景更新的问题,提出了解决方案,在NiosⅡIDE中编程实现,并下载到硬件系统中进行验证,为进一步的深入研究奠定了基础。本文将SOPC应用到整个智能视频监控设计中,具有成本低、扩展性好、携带方便及适用范围广等优点,发展前景十分广阔。更多还原

【Abstract】 With the development of science and technology and the increasing demand of public security, the intelligent monitoring system is more and more widely used in the field of industrial production, daily life, police and military affairs. The badly demand of military and civilian use of the digital video technology promotes its continuous and rapid development. Generally the images are collected by specialized image acquisition card and then processed by computer. However, because of its high costs and size, it is hard to get the extensive practical research and application, so it is necessary to design a kind of portable video surveillance system for the market.The paper introduces current state of intelligent video surveillance systems. By comparing the merits and demerits of FPGA, DSP and ARM as the core chip of the systems separately and considering the background and demands of the system, the paper puts forward the whole scheme of intelligent video surveillance technology based on SOPC. In this scheme, control logic module of peripheral interface, chips control logic module are all implemented by FPGA, and the moving object detection algorithms are realized by NiosⅡ. In order to satisfy the communication between the logic and Nios II, the paper compiles a IP core with Avalon bus, through which Nios II and peripheral device and users logic will seamlessly connect, to construct a chip programmable system.In addition, in the system of intelligent video surveillance, moving targets detection is the basic guarantee. In the paper, several moving object detection algorithms frequently-used are introduced. And by analyzing their respective advantages and disadvantages, the paper selects the subtraction background method which is easy to be realized. According to the characteristics of the subtraction background method, the paper studies the method of background extraction and background updating, proposes solutions, which is realized in Nios II IDE and downloaded to the hardware system for verification, and lay the foundation for further.In this paper, SOPC is applied to the entire intelligent video surveillance design with the advantages such as lower cost, good extension, convenience in carrying and wide range of applications, etc. Its development prospect is undoubtedly very bright.更多还原