Unsupervised video segmentation method based on feature distance

With the development of network and communication technology, large volumes of video data are generated every day. How to retrieve information efficiently from the video documents becomes a valuable issue. To address this problem, the contents of the digital videos should be annotated and indexed beforehand. Shot boundary detection is the first step for indexing and retrieving the contents from a video. Usually there exist both distinct and indistinct shot boundaries throughout a video, so the detection of these shots becomes a difficult task. According to this problem we propose an unsupervised clustering algorithm based on feature distance Io self-organize and dynamically analyse the video data, hereby segmenting the video hierarchically.