Fast content-aware video length reduction

Hsing-Ching Chang Chuan-Kai Yang

Computer Graphics & Multimedia Lab., NTUST



Video length reduction has become a very important research issue in recent years. Numerous approaches have been developed, but each has its limitations, such as fragmented results, lengthy computation time, and huge memory consumption. The main contribution of this paper is to propose a new approach that addresses all three mentioned limitations in one shot. First, by modifying the dynamic programming approach originally adopted in Avidan et al.’s work, the proposed approach extracts smooth 2D sheets and thus avoids fragmentations. Second, unlike Chen et al.’s graph-cut algorithm, this approach is much simpler and could achieve similar results but with a speed that is about two orders of magnitude faster. Third, the memory consumption is also greatly reduced to be one order of magnitude smaller. Finally and most importantly, an out-of-core scheme is also proposed, which generalizes this approach to be able to cope with videos of any lengths. Results are shown and compared with existing approaches to demonstrate the effectiveness of the proposed approach.



The result of synopsis. The left column shows original video frames and the right column shows the results of video synopsis using our algorithm. The frame number is denoted at the bottom of each frame.