Video Abstract System Based On Abnormal Events In Region Of Interest

With more and more attention to the public security, various surveillance video systems are installed in public places, which cause a large amount of video dataset every day. However, in the public security, only a small amount of video data is used for the applications such as object recognition, event detection, etc. Thus, how to find the required video data from massive video data is much desired in practice. Video summarization is such type of technology to extract the required data from a large amount of video data.In this paper, we design a video summarization system based on the abnormal event detection in regions on interest. The proposed system is very useful for obtaining the meaningful information in public security. The main content of this paper is described as follows.In the first part, we provide a literature review of video summarization from the research background and motivation. The related knowledge of video content analysis is also given in this part.In the second part, we introduce the proposed framework in detail. Basically, there are four steps in the proposed system. First, we choose the specific regions of interest in the video manually. Then we propose to use a double sliding window algorithm with double thresholds to obtain the change of cameras. Following these, the Gaussian Mixture Model is used to detect the background information and the moving objects can be detected by the differences between the foreground and background. Finally, the mean-shift algorithm is adopted to calculate the object tracking results. Through the above processes, we can summarize the video data in the specific regions of interest.In the final part of this paper, we design a component-based video summarization system. We provide the detailed introduction from the following three parts:the component construction, the video database design and the video data access. Finally, we give the implementation and demo of the proposed system. In the system design, we divide the application layer and logical layer, and thus guarantee the independence of each module and the flexibility.