Design Of Video Surveillance System Based On Cloud Computing

Nowadays, high definition (HD) video and the number of monitored channels in the security video surveillance system (SVSS) have been growing exponentially, so the ability of video data processing and storage must be suitable to this increase. Besides, video data contains large amount of abundant information, and the data size in the HD video is larger than the traditional one for several dozens of times or even more. Some special large-scale systems require a PB of storage capacity. The issue of data storage and processing is outstanding under the massive amounts of HD video data, therefore, the traditional monitoring pattern is unable to satisfy our demand. We need to resolve this bottleneck problem through the emerging technology. "Cloud computing" and "Cloud storage" technology has brought new development opportunities for video surveillance system (VSS).There are two commonly aspects of application in the VSS, one is the real-time monitoring, which requires the high clarity of video with no delay and distortion. Therefore, video data transmission speed, efficiency, security, reliability and system stability all have a very high demand. the other is history video playback, the stability of historical video data storage, the ability of the disaster backup and redundancy processing capacity will directly affects the performance of VSS. Cloud computing is a new mode of calculation, the video data transmission, processing and storage tasks are assigned to the mass cheap computers on a cluster system. According to user requirement, video surveillance applications can provide requests to the central server, such as compute, storage, and other software-related services."Cloud computing" technologies and several mainstream cloud platforms (Amazon, Microsoft Azure, Hadoop and Google) is simply discussed in this paper, After analyzing their merits and demerits, we select the Hadoop as open sources framework to build cloud platform servers. Thus computer cluster based on Hadoop framework is able to dispose PB-level data through distributed processing and storage, which reduced the cost of system development. Regarding the requirement of VSS, first we proposed a scheme based on Hadoop in cloud video surveillance system (CVSS) and then designed the system general framework, detailed project for transcoding, storage, cloud monitoring, streaming media and client will be introduced as well. Afterwards, the principles and methods are proposed since combining with the logical relationship and role in the whole system for each component. Finally, we complete the platform structures, software operation and the experimental results analysis.Consider the high real-time among tanscoding, the limited bandwidth is unable to transmit mass video data in the multichannel monitoring, the system decoding capacity is very poor. This issue prompts us to research distributed storage architecture, and we proposed a method, those two tasks about the CPU consumption and system resources transcoding, are allocated to the distributed server cluster of HDFS. Then we adopted MapReduce of programming mode, the task and parameter is submitted by client, using cluster of processing and storage capacity for real-time coding or decoding, the dynamic output to client is video streaming and must satisfy our request. This method solves a big problem in traditional VSS. Besides, in storage aspects, we introduced the child project of HDFS named HDFS distributed file technology, it means collecting the front-end video data or the transcoding data distributed deposited to cheap cluster of PC node, it is flexible to add or delete cloud node on the basis of practical request. Meanwhile, backups and redundancy can guarantee the reliability and security of video data. The storage management system based on ZooKeeper can provide the stability, because that the specific monitoring mechanisms will quickly response to the node malfunction, the processing efficiencies will not reduced by backups and redundancy. By Darwin Streaming Server development and realization of RTSP streaming server on a distributed system, multiple users access to one channel at the same time and the clarity or real-time also can be ensured effectively.This paper designed a VSS based on "cloud computing" has the advantage of distributed, big concurrent volume access storage, availability, fast response, easy to management and vindicate. The video collection, coding or transcoding, data storage and flow media distribution server, client are allocated in the physical host node which is access to the network, compared with the traditional VSS, the ability of real-time processing or transcoding efficiency even storage capacity are improved, the proposed VSS satisfy the requirement in this new period and has great of commercial value. The system function and performance tests is used in the end, experimental results indicate that the proposed VSS reduced the processing time for video transcoding, increased stability and the storage capacity.