| 摘要 |
The present invention relates to a method and apparatus for extracting key frames from a video. The method includes: dividing the video into multiple levels of video segments, wherein each video segment at a non-lowest level comprises one or more video segments at the next lower level, the multiple levels range from a first level to an N-th level, N is an integer greater than 1, the first level is the lowest level, and the N-th level is the highest level; extracting from video segments of the video at a non-highest level candidate frames at the non-highest level; for each level from a next higher level to the highest level, selecting zero, one or more candidate frames at the level from candidate frames at the next lower level in each video segment at the level; and outputting the candidate frames at the highest level as the key frames of the video.
|
