Automated real-time video surveillance summarization framework

Reviewing video surveillance contents for security monitoring is a time-consuming and time-limiting task. This paper presents a real-time video surveillance summarization framework intended for minimizing the time requirement for time critical tasks, based on compact moving objects in time-space. A tunnel is proposed as an individual time-dimension object. In order to summarize an endless video into a shorter duration without loss of selected targets so as to extend the understanding of any given individual object, this research utilizes three real-time algorithms. Direct shift collision detection (DSCD) is implemented for the extremely fast shifting of tunnels together in time-space. The DSCD summarized video can then be customized by technique from many different approaches. Here, early trajectory searching is applied with the same DSCD technique, and then direct distance transform is used to instantly give the trajectory similarity between tunnels and the user's query. The most important step for identifying each individual object is background subtraction. To this end, dynamic region adaptation (DRA) was used as the background subtraction algorithm to select the best foreground for each object before making a tunnel. DRA also helps DSCD to summarize the video more accurately. The proposed framework is able to provide the results by real-time performance approach without losing the major events of the original video stream.