Rearch On FPGA Realization Of Network Video Surveillance System Based On H.264

With the development of science and technology and public security demands increase, video surveillance system is widely used in industrial production, daily life, police and military affairs. The network video surveillance system which uses SOPC technology based on FPGA, H.264 coding technology and network transmission control technology is advanced in stability, function, cost and expandability, so the design of it has great academic and practical significance.The network video surveillance system designed in this thesis is composed of embedded image server with Nios II soft-core, correlative network equipment and a number of PC clients. Embedded image server captures real-time images and compresses them with H.264 coding algorithm and listen in network continuously. PC clients may access the server through network, and receive encoded data, then use H.264 decoding algorithm to reconstruct images and make real-time display. So the surveillant can well know the situation.During the design of embedded image server, this thesis firstly selects chip and develop kit. Secondly, the image capture subsystem is constructed and the image capture customizing module is designed with dual cache ping-pong exchange. Thirdly, the SOPC with dual Nios II architecture is designed and the means to implement customization, sharing of the memory chip, data transfer and synchronization between the dual CPUs are introduced. Meanwhile, the network server which usesμC/OS-II to implement multi-task management and scheduling is designed.The design and realization of H.264 video compression algorithm are the emphases of this thesis. In this paper, H.264 standard is analyzed, and the architectures of encoder and decoder are programmed. Then 16×16 intra prediction algorithm and macroblock scan sequence are designed, and the twice judging strategy is used for prediction mode selection. The scan sequence of 4×4 sub-block is also designed, and transformation and quantization programs are completed. The entropy coding in this paper consists of Exp-Golomb coding and CAVLC. A method based on scale judgement is designed to realize the algorithm of encoding nonzero coefficient level. Lastly, format of the bitstream which is transmitted through network is designed, and the decoding algorithm is designed too. The encoder and decoder are verified with VC++. And the test to observe the compressibility and distortion with different parameters is carried out. After algorithm verification is completed, PC client, which has functions of network access, H.264 decoding and real-time display, is designed. Simultaneously, transplant H.264 encoding program to Nios II. Finally, embedded image server and a number of PC clients are connected into network to debug and the network video surveillance system is realized.Experimental results indicate that the compressibility of this system is high and the quality of surveillance image is good, which evidently prove that the designs of software, hardware and image coding algorithm are successful. This system advances in low cost, good expandability and wide applicability, which has broad prospects for development.