Video Analysis Based on Mutual Information

In this paper we present the methods used for the analysis of video based on mutual information. We propose a novel method of abrupt cut detection and a novel objective method for measuring the quality of video. In the field of abrupt cut detection we improve the existing method based on mutual information. The novelty of our method is in combining the motion prediction and the mutual information. Our approach provides higher robustness to object and camera motion. According to the objective method for measuring the quality of video, it is based on calculation the mutual information between the frame from the original sequence and the corresponding frame from the test sequence. We compare results of the proposed method with commonly used objective methods for measuring the video quality. Results show that our method correlates with the standardized method and the distance metric, so it is possible to replace a more complex method with our simpler method.