The Software Design Of Video Surveillance System Based On HDFS

Video storage and VOD(Video On Demand) are very important to video surveillance system, and they directly determine the overall performance of the video surveillance system. With the development of cloud computing, cloud storage technology has got rapid development, and has provided a new direction for video surveillance storage system. By using a distributed storage system, it is not only to meet the growing demand for video surveillance storage devices, but also to improve the performance of video playback. In addition, it can greatly reduce the user cost by using cheap PC devices for cloud storage.This thesis develops network video surveillance system storage management module and VOD module based on HDFS(Hadoop Distributed File System). The system runs on an open-source Linux system. Firstly, on the basis of HDFS the thesis builds a distributed file system for storage; on the basis of adapter this thesis develops storage interface for using different storage devices; by using multi-threading technology, this thesis develops storage management module which has multi-task processing, which can communicate with the video surveillance platform; according to the characteristics of the video storage, develops modules including video file management and video search storage management, which meets the requirements of the storage in video surveillance platform; this thesis develops TS stream UDP push module for Darwin Streaming Server because of TCP’s low efficiency; based on Linux asynchronous I/O model this thesis develops the TS stream network transmission module; resulting from excellent control rate of UDP, develops the rate control module based on TS stream PCR field, which can improve the stability of TS stream and avoiding the network jitter. In short the storage management module and VOD module meet the requirements of video surveillance system, and improve the stability of system.