Abstract
According to the characters of abrupt change and gradual change, twice difference based method and window-difference maximum based method are proposed for abrupt change detection and gradual change detection respectively. Characters of abrupt change are revealed effectively by computing twice frame difference between two neighbor frame differences. Characters of gradual change are revealed effectively by computing non-neighbor frame difference between two non-neighbor frames and the center position of gradual change is discovered by computing non-neighbor frame difference maxima within a window. Integrating the two methods can almost detect all shot boundaries no matter what the shot boundary type is. The method does not need to set global threshold and has high adaptability. Experimental results show that the method is efficient.
Original language | English |
---|---|
Pages (from-to) | 949-953 |
Number of pages | 5 |
Journal | Beijing Ligong Daxue Xuebao/Transaction of Beijing Institute of Technology |
Volume | 25 |
Issue number | 11 |
Publication status | Published - Nov 2005 |
Keywords
- Abrupt change
- Gradual change
- Shot boundary detection
- Twice difference
- Video retrieval