Design Of Embedded Video Surveillance System Based On DaVinci Technology

Along with the development of science and technology, video surveillance system has come to the digitalized, network-based and intelligentized developing period. The appearance of DaVinci technology accelerates the process of digital video system. DaVinci technology has high-performance processor, highly integrated development tools and rich libraries. With its advantage of low cost and high flexibility, DaVinci technology can meet the demand of the innovation of real-time video products, making it can be widely used in the digital Video area. Based on this technology, the realization of embedded digital video systems is discussed.The system combines of TI high-performance digital media DSP TMS320DM6437’s powerful computing capability and Samsung ARM9S3C2440’s system control to achieve dual architecture. The design is practical and scalable.In order to make data transmission stable and efficient, ARM communicates with DSP through the HPI(Host post interface). The HPI hardware connection design is given and the driver of HPI under Linux runs on ARM to realize transmitting video data.System is divided into video processing module based on DSP and video transmission module based on ARM. The two modules developed independently. The process of video capture, encoding and transmission is described in detail. DSP-side video processing module main achieved the video capture and encoding. Programming on the DSP side is in TI’s integrated development tools CCS. Running on the TI DSP real-time operating system kernel, DSP/BIOS, the module called Codec Engine to achieve H.264video encoding. Video transmission is in accordance with RTP protocol standards. The open source RTP library jrtplib is transplanted on ARM to realize transmitting.Due to the uncertainty of Internet bandwidth, a QoS control strategy is given in order to ensure video transmission. The strategy first calculates the packet loss rate according to the information RTCP protocol packet provided and then adjusts the encoding rate. The rate is increasing when the network is light-load and the rate decreases rapidly when the network is high-load.Finally, a client is provided to receive the video. The client is written with C++and the interface is prepared with QT4. The video transmission module is tested and the experimental data is given as well as the images.