The Quality Scoring And Fault Detection System Of Video Surveillance

The Technology of Video Surveillance has improvement for many years and becoming more and more important in our daily life. With the time goes on, the range of the usage of this technology is also become broader and broader especially in the fields of financial, transportation, military and something other about the national economy and the people’s livelihood. But with the development of science and technology, especially the tremendous advance of network technology, there are more and more novel usage pattern integrated with the internet technology emerged recently and bring about many new problems and challenges about this traditional technology. Among these new problems, the loss of video quality come up with the internet environment’s instability is becoming the most urgent problem should be solved.After deployed in the internet, the Video Surveillance System will be faced with not only those traditional problems such as sheltered, skewing but also some new problems, such as mosaic, black frame, freezing frame, brought out by the instability of network situation. The quality of the video will also be damaged at the same time. Because of the giant quantity and the variety of the kinds of video devices, detecting these problem by artificial monitoring is impossible and it is very necessary to develop a new Automatic Video Surveillance System Fault monitoring System which is used to automatically detect the fault of Video Surveillance System and raise the alarmThe Automatic Video Surveillance System Fault monitoring System is consisted with the user layer video quality evaluation part and network layer video quality evaluation part. The user layer part is mainly analyzing the video content and the network layer is mainly get the MOS value by collecting the video encoding information, such as bitrate and frame rate, and the network situation, such as packet loss rate, as the input to get the MOS value. The introduction of the network layer is the main point of this paper.The main design method of the network layer is collecting the transcendental MOS value in different encoding situation and network environment as the training data set to get the prediction model. This method is mainly reference the IPTV quality prediction model which can get the value without reference. In order to introduce these method into the Video Surveillance System, I propose some amend method to make the model more suitable for our system.At the same time, after abundant experiment, we research how did the instability and the network fault influence the equality of the video. In another word, what would happen when the network affected by the different packet loss rate. Using the statistical method, we just get the preliminary experiment data, and I think it will make a contribution to the further research and the system optimizing.Our system framework is based on the J2EE and database. User can use web browser to get the video quality information from our system, and user can also get the information from the database by different user platform which can ensure the user can know the system situation no matter when, where and how to get it.