| 摘要 |
Browsing videos quickly is a common way to check their contents. It is desirable to create automatically an abstract of a video sequence in a page by page story book, so that the temporal order among images is preserved strictly and can easily be identified. A method for generating and displaying a video abstract comprises separating a video into a plurality of video shots, determining within each video shot a keyframe, extracting from the keyframes salient regions (ROI) and providing control signals for displaying the salient regions on a display page. Particularly, the ROIs may be resized, and their locations are determined according to the sequential order, wherein vertical boundaries in different lines are disaligned (D) in order to improve the intuitive recognition of the temporal structure.
|
