Jun Su Kim, Deokmin Haam, and Myoung Ho Kim
Key frame extraction, Representative frame, Skyline operation, Video abstraction
Key frame extraction is the technique for extracting a small number of frames that best reflect the contents of a video. In general, many key frame extraction methods use multiple features in a target video, and calculate multiple scores of a frame based on each feature space. Then they combine the scores to calculate the final score of the frame by making their own weight functions on those scores, where the weights determine relative importance among features. Those methods using the weights of features have some difficulties caused by the weights. First, since many methods set the weights by their experimental results, the weights are needed to be determined again when the used features or target videos are changed. Second, some important frames get lost a chance to be key frames by the weights of features. In this paper, we propose a novel key frame extraction algorithm, which applies a skyline operation to find key frames from each shot. By using a skyline operation, our method can extract key frames without the difficulties caused by the weights of features. The experimental results show that our method can extract key frames from videos effectively.
Important Links:
Go Back
