A Key Frame Based Video Summarization using Color Features

Abstract

The rapid development of digital video has opened ways for several multimedia applications. A huge amount of video data is uploaded on internet on daily basis, which, consumes additional internet bandwidth and storage space. This rapid growth of video data requires an effective and user-friendly content summarization method that can efficiently browse the video contents. Different from recently proposed approaches, we present a video summarization method to detect shot boundaries based on a combination of color features extracted form patches of a video frame instead of a whole frame. This makes the proposed approach more robust to the types of video transitions by accurately detecting the video shots. Each video shot is further divided into subshots by comparing the structural similarity among the frames, to extract a keyframe from the most representing subshot of an individual video shot. Finally, the extracted keyframes from the subshot of each video shot, are compared individually to remove the redundant frames. Experimental results ofthe proposed approach on videos extracted from Open video dataset, confirm its effectiveness, by comparing with the state-of-the-art techniques.