Information and Entropy Measurements on Video Sequences

Video compression techniques usually involve lossy compression algorithms, with much of the original image data being discarded because it is not essential for human perception. However, lossy compression not only reduces data volume substantially, it also removes a reference for how well the remaining data has been compressed. Entropy measurements can go some way towards answering this questions, but have traditionally been difficult to perform. This paper uses a novel O(N log N) entropy measure to investigate video frame sequences from the standard MPEG-2 file set.