Design Of Embedded Network Video Surveillance Encoder And Research Of System Real-time Performance

In recent years, video surveillance system has been widely used in various vital industry, such as finance, posts and telecommunication, electric power, traffic, business, tourism, public security, military, medicare, mining and so on. Research on networked video surveillance system has been considered be of significance, because of wide application and exigent requirement on video surveillance system.We design and implement an embedded network video surveillance encoder for different applications. MPEG-4ASIC is adopted in the encoder, based on the available ARM9system. In the system, analog video and audio signal are converted to digital data and compressed by MPEG-4encoder, and then the encoded video and audio data is transported to the monitor terminal over network. Finally, the supervisor of the system can see the video and hear the audio transferred by the video encoder, so he or she can monitor the whole process. We have finished the scheme design of video and audio part, device driver based on ARM linux. Furthermore, we have implemented synchronous transformation of video and audio using RTSP protocol. In this paper, we have done thorough research on the real-time performance of embedded video surveillance system, apart from implementation of fundamental function of embedded video surveillance system. We propose a scheme to enhance the real-time performance of embedded network video surveillance system and to minimize memory size based on the existing scheme. In order to achieve the target, the scheme adopts the linear increment video traffic prediction algorithm and uses the packet pair to detect the available network bandwidth. Furthermore, we propose a memory allocation algorithm to save memory resources.The embedded video surveillance encoder, designed in this paper, has excellent real-time performance and video quality performance. The delay of the system is less than500ms in local area network and the real-time performance of the system has come up to domestic leading level. The image of the system is clear and continuous. Using no-reference image quality assessment method to evaluate the image quality of our system, its image PSNR is more than39db.