Research On Efficient Coding Technology Of Network Video Monitoring System

It provides a good technology supporting platform for the video monitoring technology with the rapid development of the computer technology, information technology, internet technology and the computer vision technology. From the Original analog video monitoring to current digital network video monitoring, people can find the video monitoring technology is to mature. To today's network video monitoring, people can navigating the scene by the IP host no matter what time and where he is in. However, there are many problem to overcome in the process of researching and development of the network video monitoring. Due to the limited network bandwidth and the amount of the video code stream data captured from the camara, the video code stream data transport to the network like a big truck travel on the narrow road. So it can't be transported correctly and rapidly. For this reason, it can reach the requirement to transport in real time. Therefore, to transport the video code stream in a small quantity and more quickly and its vision quality can't be influenced as well, many experts do a lot of analysis and reaserch about the video sequence. They made use of every techonology to compress video data. The most important compressed technology is using the relevance between pixel and pixel or between video frame and video frame. they were named Intra-Prediction and Inter-Prediction.Because of the network video monitoring system can apply to different network which using different network protocol, to compress the video data in most possible, the author do a research on the video data compression coding according to the portability and high compression ratio of H.264/AVC.To transport video data in less code stream and assure the client can see the video clearly, the author did an analysis and research on the existing moving object detecting technology in the the thesis at the first. In the thesis, three moving object detection methods were listed in static scene. they are optical flow, neighbour frame difference algorithm and background subtraction. The thesis made a compare among these methods and chose the background subtraction method to detect the moving object. Because the background subtraction method only detect the static scene, the thesis present a new method which can detect object in dynamic scene, rapid adaptive updating background. To study the moving object detecting technology further, the thesis make an analysis on the motion estimation and compensation combining video coding frame of H.264/AVC. On the basis of research, the author also made a detail study on the motion estimation method based on block matching. Because there is a misjudgment in the SAD in the situation of the monitoring scene light changes, the author presented a new matching criterion,N_SAD. At last, the thesis took an advantage of the OpenCV which supported by Intel to realize the object detecting system under the environment of vc++6.0.