Design And Implementation Of Realtime Panoramic Video Compression And Transmission System Based On H.264

With the rapid development of information technology and multimedia technology, video surveillance application scenarios become more widely. Traditional local video surveillance cannot satisfy people’s needs, the demand for remote video surveillance are increasingly strong. While video contains amount of data and our country’s network bandwidth is limited, in the field of remote video surveillance, especially for the general aviation airport remote video surveillance, a large number of video data would be transmitted in the limited bandwidth, in order to improve surveillance and reduce the accident rate, real-time panoramic video compression and transmission technology has become a hot topic and difficulty. In this paper, Miyun general aviation airport remote video surveillance application as the background, the real-time panoramic video compression and transmission techniques are studied, designed and implemented the realtime panoramic video compression and transmission systems based on H.264.The main work and innovation of this paper are as follows:1. The requirements of real-time panoramic video compression transmission system is analyzed, and the requirements summarized clearly in the form of use case diagram, then the key issues and non-functional requirements was summarized.2. For existing front-end multi-channel video capture and backend panoramic video stitching program exists the problem that multiple video images are not synchronized and higher bandwidth is required. This paper proposes a system architecture that stitching in the front-end and watching in the back-end, effectively achieves the multichannel video images simultaneously, and can reduce the redundant data transmission.3. For video stitching takes a lot of time, this paper uses the offline thread to compute transformation model, online real thread to stitch, and use CUDA technology to accelerate. For video compression problem, we use the X264 encoder to encode the video, and optimizes encoding parameters.4. For the higher need for real-time, this paper use modular design solution, the whole system is divided by main functions, each module runs in a separate thread, achieve that the video capture, mosaic, compression, transmission run in pipeline, and bind each thread to a different CPU core to further optimize multi-threaded program, and finally the system latency is controlled within 800 milliseconds.