Keyframe extraction based on kmeas results to adjacent DC images similarity

Keyframe extraction is the fundamental process of video content analysis, retrieval and so on. For extracting keyframe from video compressed stream efficiently, this paper presents an useful and fast method. It firstly computes the similarity set of adjacent I frames' DC images, secondly uses kmeans algorithm to cluster the similarity set, and finally selects keyframes based on the clustering results. The experiment results show that our method is able to get proper keyframes from test video files and finish keyframe extraction in much less time.