Shot-Change Detections and Shot-Change Effects Recognition based on Motions and Their Estimation Reliabilities

There are many shot-change detection methods. They work well in videos without large motions and fast camera works. This paper proposes more robust method for detecting shot changes and recognizing their shot-change effects. The method uses the motion between frames and their estimation reliabilities. With motion estimation, we can have the correspondence between frames. But we have false estimation when there is a shot-change between frames. The proposed method estimates the reliabilities of the motion estimation. With the reliabilities, the proposed method estimates the probabilities of correspondence between frames. This reliability and the estimated motions lead the estimation of existence of shot-changes and the type of shot-change effects. First, this paper proposes the shot-change detection method based on the motions and their estimation reliabilities. Then we show the experiments on simulated videos and real videos. And last, we conclude our work.