Animation Movie Abstraction: Key Frame Adaptative Selection Based on Color Histogram Filtering
Laurent Ott, Patrick Lambert, Bogdan Emanuel Ionescu, Didier Coquin · 2007
This paper deals with the construction of animation movie abstracts. The proposed video abstracts are computed using an adaptive selection of key frames obtained by color histogram analysis. The method exhibits three main parts. In the first step, a shot detection is performed. Then, within each shot, an adaptive number of key frames is selected. This selection is performed by analyzing cumulative distances between frame color histograms. In the third step, an inter-shot examination is performed. Using an iterative selection we provide a user-specified number of movie representative frames. The method has been tested on a database from the International Animated Film Festival that takes place yearly in Annecy.