Research And Implementation Of Cloud Computing Based Network Video Surveillance System

Network surveillance system based on cloud computing architecture can not only solve the problems faced by traditional network video surveillance system, such as the huge amount of data transmission, shortage of resources for network bandwidth, mass storage for video files, and inadequate computing capacity, but also fundamentally guarantees the security, stability and effectiveness of the system. It makes better use of the cloud storage infrastructure to optimize data access approach, using cloud computing to distribute the complex tasks to the cloud and complete them, as well as integrating both wired and wireless network surveillance into the same cloud. Therefore, the study of cloud computing based network video surveillance system technology is of great significance.The proposal of VSaaS makes it more convenient to develop the network video surveillance system and leaves it more available to use. VSaaS has a significant impact on cost control of the network video surveillance system since the system architecture may buy the computing capacity based on its own demands. Considering the facts of the data security, low demand for the client, easy data sharing and unlimited ability to expand, the cloud computing effectively responds to demands of network video surveillance for real-time, security and data sharing.Real-time surveillance through the network transmission is very concerned about network communication and system storage capacity. In this paper, the research of the visual surveillance technology of the traditional network video surveillance system is deeply researched. To effectively utilize the bandwidth and reduce the amount of video memory, an improved method of video motion detection is provided and it has been used in an actual project.In this paper, the author analyzes the three core technologies of Google Cloud, GFS, Map-Reduce and BigTable, and researches the specific developing deployment of cloud computing systems. Moreover, such research has been employed in the actual project, An Yu video surveillance system on the GAE-based video surveillance system Structure, in which based on GAE in the cloud, three major functions, video data storage, video communications and processing data calculation, have been realized by Python and JAVA programming, and DirectX technology is utilized in the acquisition module to capture video data.Finally, targeting at key technologies and difficulties, such as system deployment, motion detection, video transmission, mobility support and others, of the cloud computing based network video surveillance system, this paper proposes the implementation scheme, and introduces the design and implementation of both hardware and software of the system.Through the architecture and implementation of the actual system studied from the feedback of the result, the author draws the conclusion that the development of video surveillance in this area need to break in two directions, one is cloud based computing, the other is intelligent. To intelligently analyze and process the video data in this system, the moving target detection has just served as a module test applied and deployed in the surveillance cloud. However, the experiment reflects the good results. In the future, technologies of face recognition, behavior analysis, characteristic analysis and others will surely play an important role in the intellect surveillance by means of blade service so as to be easily integrated into the system. And the other major work is to deploy the traditional video surveillance system into the Google cloud. According to the market feedback, compared with the previous generations of products, such method reduces costs, facilitates management and improves the stability and ease of expansion.The main characteristic of the system are as follows: to realize parallel computing by Map-Reduce programming through distributing the complex video data computing to each leaf node so as to solve the traditional problem for the lack of monitoring system capacity, and to introduce the intelligent identification technology into the video surveillance system by the implementation of the improved algorithm of moving target detection and get good result in the experiment. Besides, the author changes the traditional mode of video surveillance system from C/S into B/S, so that all packets follows the http communication protocol and a true cross-Platform is realized.