“Multi-frame video representation using feature preserving directional blur”

Conference:


Type:


Title:

    Multi-frame video representation using feature preserving directional blur

Abstract:


    The long-standing goal of the video-based graphics community is to be able to view video content at one view. This sketch shows a new video representation method for visualizing video content quickly and effectively. Traditionally, a video thumbnail or movie thumbnail is used for a video browsing interface. It is insufficient to represent the video content because it shows only one sample frame of the video or it is annoying for viewing because of the difficulty of distinguishing the beginning and the end frame of the movie content. On the other hand, a multi-frame representation is used for editing and quick viewing of the video sequence [1]. However, it is hard to view when each frame overlaps and a special frame alignment is necessary to compose a summarized frame. We propose a “video flash” representation method, which uses transparent multi-frames with directional blur effects dependent on a layout. A directional blur effect is used to diminish noticeable artifacts on overlapping frames and visualize frame sequences at the same time. Furthermore, there is no need to consider each frame alignment in detail. This method is also seamlessly transferable from conventional movie thumbnails and applicable for new kinds of video browsing interfaces with a quick preview mode.

References:


    1. Teodosio, L. et al., Salient Stills, ACM Transaction on Multimedia Computing, Communications, and Applications, Volume 1, Issue 1, pp. 16–36 (2005).
    2. Brostow, J. G. et al., Image-based motion blur for stop motion animation. Proc. of SIGGRAPH’01 pp. 561–566 (2001).


ACM Digital Library Publication:



Overview Page: