A Novel Shot Detection Algorithm Based on Clustering

摘要

Shot boundary detection has attracted much more research interesting in recent years. This paper present a novel shot boundary detection algorithm based on K-means clustering. At first the feature of color is extracted, the dissimilarity of video frames is defined.Then the video frames are divided into several different sub-clusters through performing Kmeans clustering. It can detect cut and gradual shot by the adaptive double threshold of different subclusters. The efficiency of the proposed algorithm is extensively tested on movie, news and other videos. The experiments results indicate the method had a high accurate rate in both cut shot detection and gradual shot detection.