Implementation And Research On Quick Video Browsing System Algorithm Based On Sparse Moving Objects

With the rapid economic growth, rapid social progress and continuous improvement of people's lives, the requirements of the public on public safety and security systems is increasing and the demand for public safety and security systems grows highly, thus promoting the development of video surveillance systems. There is a great significance on investigation into the events by video surveillance system records. When an abnormal situation occurred we need to access to and playback the original video. However, the original video data saved as little a few days then more than a few months. To find the interesting object accurately from the mass of video data, it will take a lot of time and effort on playback the original video through the traditional method. The process is boring, likely to cause mental fatigue even lead to negligence and mistakes. Therefore, it is necessary to find a method for browsing the objects in the mass of video data quickly.In the paper,The main contents researched as follows:1. The quick video browsing method achieved in this paper mainly using motion detection and segmentation, object tracking technology. Therefore, the common object detection and object tracking methods were introduced, experiments were done to compare and analysis the result. Pointed out the advantages and disadvantages of each method and the range of application.2. In this paper SURF features is used for object tracking, the object location is determined by matching the SURF features of the template image and each sub-image containing moving object in video sequence. It is a local feature used for object matching in the paper, with the scale and rotation invariance. The features have very good adaptability for zooming, small perspective changes, noise, and brightness changes. Experimental result shows that the method is a simple and efficient object tracking and identification method.3. In this paper, moving object detection and segmentation, object tracking and other technology is used to achieve a quick video browsing method. And implementation of dynamic control of the moving object density. Classification of video segment by the length of the video and the balance of the moving object density are resolved. dynamic control of the moving object density means that controlling the moving object density in the composite video interactive by statistics the density of moving object segmented in original video. Avoid of that the moving objects are too dense to distinguish each other. and the moving objects are too sparse may cause increase the length of the video.