基于TMS320DM368的智能高清监控系统的研究与实现

发布时间：2018-12-20 14:21

【摘要】：视频监控技术在安防领域发挥着极其重要的作用。随着社会的发展,传统的视频监控已经不能满足于市场需求。智能监控作为监控领域最前沿的应用模式之一,具有响应速度快、主动性强、可靠性高、用途广等优点,因此有很重要的研究意义。本文通过研究视频监控的基本原理、高清数据的采集处理、目标跟踪的技术成果,设计并实现了一个基于TMS320DM368平台的智能高清监控系统,该系统可从画面分辨率和实际使用性两个方面改善监控效果。论文首先完成了系统的硬件设计,具体包括DM368平台、MT9P031图像传感器和PTS-301Q云台。其次研究了DM368的特点、MT9P031采集视频的基本原理、V4L2驱动获取数据的方法、H.264压缩编码原理以及网络传输协议TCP/IP,并编程实现了高清视频的网络传输。在此基础上,本文研究了运动目标跟踪算法。利用三帧差分法检测出运动目标后,用改进的Camshift算法对目标进行跟踪。从三个方面对Camshift算法进行改进：只对目标的局部背景区域进行反向投影计算；计算目标与干扰区域的颜色相似系数,当系数大于阈值时,采用H-S二维直方图生成反向投影图；当目标被遮挡或突然消失时,结合Kalman滤波预测目标的运动状态。最后,文中通过一个简单、有效的算法控制云台转动,可使目标始终位于监控画面中心。测试结果表明,该系统可支持分辨率为1280×720的高清视频网络传输,能够实现准确的目标跟踪,提高了系统的监控效率。
[Abstract]:Video surveillance technology plays an extremely important role in the field of security. With the development of society, the traditional video surveillance can not meet the market demand. As one of the most advanced application modes in the field of monitoring, intelligent monitoring has the advantages of fast response, strong initiative, high reliability and wide application, so it has very important research significance. Based on the research of the basic principle of video surveillance, the acquisition and processing of high-definition data and the technical achievements of target tracking, an intelligent high-definition monitoring system based on TMS320DM368 platform is designed and implemented in this paper. The system can improve the monitoring effect from two aspects: picture resolution and practical usage. Firstly, the hardware design of the system is completed, including DM368 platform, MT9P031 image sensor and PTS-301Q platform. Secondly, the characteristics of DM368, the basic principle of MT9P031 video acquisition, the method of data acquisition by V4L2 driver, the compression coding principle of H.264 and the network transmission protocol TCP/IP, are studied. The network transmission of high-definition video is realized by programming. On this basis, the moving target tracking algorithm is studied in this paper. After the moving target is detected by three frame difference method, the improved Camshift algorithm is used to track the target. The Camshift algorithm is improved from three aspects: only the backprojection of the local background region of the target is carried out; The color similarity coefficient between target and interference region is calculated. When the coefficient is larger than the threshold value, the H-S two-dimensional histogram is used to generate the reverse projection image, and when the target is blocked or suddenly disappeared, the moving state of the target is predicted with Kalman filter. Finally, a simple and effective algorithm is used to control the rotation of the cloud head, so that the target can always be located in the center of the monitoring screen. The test results show that the system can support high-definition video transmission with a resolution of 1280 脳 720. it can realize accurate target tracking and improve the efficiency of system monitoring.
【学位授予单位】：南京理工大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TN948.6

【参考文献】