Design And Implementation Of Image Transmission Adn Distribution System For Real-Time Video Surveillance

In recent years, the technologies of intelligent video surveillance based on image processing have become a hot topic in the field of computer vision and security. Generally, we use H.264and RTSP streaming media technology to compress and distribute the image data generated by the intelligent video surveillance to the network. But the process of H.264encoding costs much cpu resources, and the process of decoding has a few seconds delay, so it is unable to achieve real-time video surveillance in the browser.This paper studies the representation and compression of image, and analyzes the transmission of image data and the real-time communication technologies between browser and server. Then we design and implement an image transmission and distribution system based on Node.js. This system can receive multiple JPEG data streams sent from the intelligent video surveillance system based on TCP protocol, and distribute them to the network based on WebSocket protocol. In the browser, we use WebSocket to recive the image data and render it fast based on object URL. The system realises that monitoring staffs can perform remote and real-time video surveillance in the browser. Furthermore, we propose a kind of network congestion control mechanism based on transmission delay feedback and camera weights. This mechanism grades the congestion based on transmission round trip delay, and adjusts the transmission rate of image data based on the congestion grade and camera weight. Then it can ease the network congestion and maintain the effect of real-time video surveillance at the same time. At last, we experiment on the transmission delay, the frame rate, the quality of image, the transmission rate, and the server performance to prove the feasibility and the reliability of the system.