当前位置:主页 > 科技论文 > 网络通信论文 >

基于HDFS的视频监控系统软件设计

发布时间:2018-06-08 11:36

  本文选题:视频监控 + HDFS ; 参考:《浙江大学》2014年硕士论文


【摘要】:视频监控系统最重要的是视频数据的存储和视频的点播回放,同时视频存储设备作为视频监控系统的基础设施,存储设备的稳定性及性能直接决定了视频监控系统的综合性能。随着云计算的发展,云存储技术也得到了快速发展,同时为视频监控系统的存储提供了新思路。通过引入分布式的存储系统,不仅可以满足日益发展的视频监控系统存储设备的需求,同时能够提高视频回放的性能。另外由于云存储技术支持使用廉价PC设备搭建分布式集群,可以大大节约用户成本。 本文基于HDFS(Hadoop Distributed File System)分布式文件系统研发了网络视频监控系统存储管理及VOD回放模块。本文研究的系统运行于开源Linux系统上,将开源的分布式文件系统HDFS应用于视频数据的存储和读取,研究开发了基于HDFS集群的存储管理和VOD点播模块;研究了HDFS分布式文件系统的架构及关键技术,并基于HDFS构建了分布式存储集群;为了能够对不同存储进行适配接入,应用适配器模式开发了存储接口用于存储设备的接入;应用多线程技术,基于多任务处理的方式研发了存储管理模块用于接入视频监控平台;根据视频存储的特点,开发了包含录像文件管理、录像搜索的存储管理模块,满足了视频监控平台存储模块的要求;针对传统TS码流TCP over RTP传送效率低的缺点,基于Darwin Streaming Server研发了TS码流UDP推送模块,提高了系统性能,增加了DSS的系统功能;基于Linux异步I/O模型,研发了TS码流网络传输模块,提高了系统的性能和稳定性;由于UDP没有良好的速率控制,基于TS码流的PCR字段开发了速率控制模块,提高了推送TS流的稳定性,有效避免了网络抖动给客户端带来的马赛克或者延时影响。所研发的存储管理模块和VOD回放模块,满足了监控系统的存储和回放要求,并且提高了监控系统的稳定性。
[Abstract]:The storage of video data and the playback of video on demand are the most important parts of the video surveillance system. The stability and performance of the storage device directly determine the comprehensive performance of the video surveillance system. With the development of cloud computing, cloud storage technology has also been rapidly developed, and provides a new idea for video surveillance system storage. By introducing the distributed storage system, it can not only meet the requirements of the increasingly developed video surveillance system storage devices, but also improve the performance of video playback. In addition, the cloud storage technology supports the use of cheap PC devices to build a distributed cluster, which can greatly reduce the cost of users. In this paper, a storage management and VOD playback module of network video surveillance system is developed based on HDFSX Hadoop distributed File system. The system studied in this paper is running on the open source Linux system. The open source distributed file system HDFS is applied to the storage and reading of video data. The storage management and VOD on-demand module based on HDFS cluster are developed. The architecture and key technologies of HDFS distributed file system are studied, and the distributed storage cluster is constructed based on HDFS. Based on multithread technology, a storage management module is developed for accessing video monitoring platform based on multi-task processing, and a storage management module including video file management and video search is developed according to the characteristics of video storage. It meets the requirements of video monitoring platform storage module, aiming at the low efficiency of traditional TS stream TCP over, a TS stream UDP push module based on Darwin streaming Server is developed, which improves the system performance and increases the system function. Based on the asynchronous I / O model of Linux, the TS bitstream network transmission module is developed, which improves the performance and stability of the system, and because UDP does not have good rate control, the rate control module is developed based on the PCR field of TS stream. The stability of push TS stream is improved, and the mosaic or delay effect of network jitter is avoided. The developed storage management module and VOD playback module meet the storage and playback requirements of the monitoring system and improve the stability of the monitoring system.
【学位授予单位】:浙江大学
【学位级别】:硕士
【学位授予年份】:2014
【分类号】:TN948.6

【参考文献】

相关期刊论文 前5条

1 李炳林;流媒体技术及应用[J];电力系统自动化;2001年24期

2 许静,陈宏刚,王庆人;软件测试方法简述与展望[J];计算机工程与应用;2003年13期

3 黄拔峰,钟明,杨传钧,张家钰;Darwin Streaming server的研究与应用[J];计算机工程;2004年19期

4 刘运强,王汇源;Socket和多线程在视频传输中的应用[J];山东大学学报(工学版);2004年02期

5 卢静;刘占伟;;基于H.264的嵌入式视频服务器的研究与实现[J];信息与电脑(理论版);2010年01期

相关博士学位论文 前2条

1 肖广;基于IP的视频监控若干关键技术研究[D];华东师范大学;2008年

2 黄源源;视频监控系统中一些关键技术的研究[D];电子科技大学;2013年

相关硕士学位论文 前2条

1 余少勇;基于DM642的流式网络视频监控系统研究[D];厦门大学;2007年

2 张兴;基于Hadoop的云存储平台的研究与实现[D];电子科技大学;2013年



本文编号:1995656

资料下载
论文发表

本文链接:https://www.wllwen.com/kejilunwen/wltx/1995656.html


Copyright(c)文论论文网All Rights Reserved | 网站地图 |

版权申明:资料由用户66ad8***提供,本站仅收录摘要或目录,作者需要删除请E-mail邮箱bigeng88@qq.com