Content Based Unsupervised Video Summarization Using Birds Foraging Search

摘要

Video summarization is the process of making summary of a video by extracting the key frames. Our content based video summarization approach is based on birds foraging search optimization algorithm. In this proposed method, windowing rule is used to retrieve the candidate frames of the video. Contours of different angles with the horizontal axis are detected from several blocks of the candidate frames. We group the blocks according to the angles. Color moments are used as color features of each group. This contour group based features represent contents of the candidate frames. It is suitable because of its ability to group the features of the same portions of an object irrespective of its location. We adapt population-based birds foraging search optimization algorithm to determine the key frames from the candidate frames. It is evident from the experiment that our proposed method produce better result than some related works.