【作者】 郭静；

【导师】 褚晶辉；

【作者基本信息】 天津大学 ， 信息与通信工程， 2012， 硕士

【摘要】 网络技术和多媒体技术的突飞猛进带动了安防产业的快速发展,作为其重要手段的视频监控凭借其方便、触角广等优势已渗透到了很多行业,并经历了从模拟到数字的历史转变,朝着网络化、智能化和信息化方向发展。目前如何在网络上有效实时传输监控信息是业内普遍关注的热点,而智能视频监控作为监控领域中最前沿的应用模式之一,其研究受到了多方重视,尤其是学术界和产业界。在该背景下,本课题基于微软公司的DirectShow技术实现了一个具有智能跟踪识别功能的网络视频监控系统。DirectShow技术是优秀的流媒体处理技术,解脱开发人员于底层硬件细节而使其专注于上层逻辑的设计,使开发变得简单而高效。本文的研究以DirectShow技术为平台,在充分进行了需求分析的基础上,对系统设计过程中涉及的关键技术进行了剖析,提出了系统的总体设计,并以此为导向进行了详细设计,主要完成了系统平台搭建和软件界面设计实现,涉及单路监控链路的实现、多路监控、抓帧Filter的加入、嵌入智能跟踪识别算法和日志等模块的设计工作。通过对典型网络视频监控系统的改进,开发出了一套高效率和模块化设计的智能监控系统,实现对多路远程IP摄像头的实时监控,同时还解决了对监控视频中出现的异常情况进行重点跟踪与识别、镜头切换跟踪等问题,使监控功能更加强大。论文还总结了系统开发过程中的要点和问题,对系统性能进行了测试,基本实现了所需要求,同时对系统的改进提出了方向和建议。更多还原

【Abstract】 The technological advancement of network and multimedia leads to the rapid development of the security industry. As an important means of security industry, video surveillance, with its easy and wide reach advantages, has penetrated into many fields, and is undergoing the historical transformation from analog to digit and going towards network-oriented, intelligent and information-based. How to make effective transmission of the real-time monitoring information through the network is the main concern of the industry in general. Intelligent video surveillance, as one of the most frontier application mode, its research has drawn attention from many fields, especially from academic and industry. Under such backgroud, this thesis implements a networked video surveillance system with the function of intelligent recognition and tracking based on DirectShow technology provided by Microsoft company.DirectShow is an excellent technology for streaming media processing, and can free application developers from hardware details and make them focus on the upper logic design. It makes the application development simple and efficient. The research in this paper, on the platform of DirectShow, carried out fully needs analysis, involved deep research of key technologies and at last proposed the general system design and the detailed system design. The major project work is about the system platform building and the software UI design and realization, including single channel filter chain building, multi-channel extention, Sample Grabber filter joining, the intelligent recognition and tracking algorithm embedding and the log identification module implemention. The system fulfilled in this thesis is of great efficiency, modularization and power compared to typical surveillance system. It achieves real-time surveillance using multipath remote IP cameras and can handle issues of motion tracking and recognition as well as tracking with viewpoint shift, etc. The thesis also summarized the key points during the development. The system performance was tested and reached the requirements and further improvement advice and direction were suggested to the system.更多还原