Computational Video Editing Model based on Optimization with Constraints Satisfaction

摘要

This paper presents a computational video editing model. In this model, a video is considered as a sequence of small boxes each of which has a length of 0.5 or 1 second, and the editing is defined as the problem of filling each box with an appropriate shot. Each editing rule is implemented independently of the others, as an evaluation function or a constraint for choosing shots and arranging shots. This formalism enables easy and systematic editing investigation by including or excluding editing rules. Based on this model, our objective is to support for the generation of a variety of editing patterns or a simulation of typical editing in movies and TV programs. We implemented this model in an actual video editing system, and we verified that the model works for multi-angle videos, and that our set of rules produce good results.