Research And Implementation Of Video Summarization Based On Monitoring

With the rapid development of microelectronic technology and computer technology, mass data such as videos accumulates constantly. It is more and more difficult for the user to search for and process the videos. Thereby, people have to explore and improve the videos data processing technology. Nowadays, the video summarization technology related to video processing, application of monitoring technology in particular, plays a more and more role in all industries, which greatly decrease the workload of staff.As it is nonlinear structure, video data cannot be processed by common method. Video summarization technology is one of the various methods to solve this problem.The context mainly researched the video summarization technology in video processing. The major procedures in video summarization include video reading, background modeling, foreground extracting, moving object tracking, temporal and spatial programming and generation of concentrated video. The context introduced the core technology in those procedures in details. On one hand, the context put forward improved algorithm in object detection background model building. On the other hand, the context design a video summarization model system based on improved algorithm.At first, the context researched the moving object detection algorithm and divided it into four types according to the basic principles, simple background modeling, statistical information modeling, the non parametric kernel density estimation modeling and non background modeling. In addition, the context chose typical algorithm to introduce and test each model. When using frame differential method to extract foreground, the context put forward one improved algorithm, which can abate the coarse image problem effectively.Secondly, the context researched the object tracking methods. Moving object tracking is the optimizing technology in the whole design. If used properly, it can make the system more intellectualizing and humanizing. The charter introduced the UKF, Mean Shift methods and give basic principles of algorithm.According to video summarization technology, building background model and movingaverage method based on average background method, the context put forward, oneimproved algorithm based on traditional moving average algorithm and carried ouexperiment to analyze the algorithm.At last, based on vs2010 platform, the context design and explore video summarization system, gave major system module and flow diagram and made analysis according to the actual monitoring video test result.