A Feature Distance based Algorithm for Video Segmentation

Z.-H. Sun, P. Fu, J. Xiao (PRC), and D. Meng (Australia)


shot boundary detection, indistinct shot boundary, FDFCM


As the rapid development of the use of Internet, people have more experience to access the rich information. However the mess status of the media organization usually discourages us. It's obliged to build a sound structure for the media data. In this paper we solved the problem of building structure for video sequence. For the convenience of retrieving information from video database, the contents of the digital videos should be annotated and indexed before being stored in the database. Shot boundary detection is the first step for labeling and indexing the contents of a video. Since there exist some indistinct shot boundaries throughout a video, the detection of these shots becomes a hard work. According to this problem we propose an unsupervised fuzzy clustering algorithm based on feature distance (FDBFCM) to self-organize and dynamically analyze the video data, hereby segmenting the video hierarchically.

Important Links:

Go Back