Picture coding standards have been developed and are now in position to be practically the most effective methods. However, some redundancy is found as a certain correlation for longer interval on coded bitstreams of pictures[1-5]. Entropy derived from the probability of occurrence of fixed length bit patterns in coded bitstream is investigated. A small redundancy is found for the interval with a length up to 20 bits in the tested bitstreams. Entropy for the interval with length of 30 bits on coded bitstream is difficult to be calculated because the data size exceeds available memory capacity. To reduce data size, the authors calculate entropy for sampled patterns based on an assumption that the optimum coded bitstream is a random sequence. Based on this assumption, we obtained an necessary length of coded bitstream for accuracy analysis. The analysis shows that the entropy decreases for some bitstreams, which indicates that re-encoding algorithm for bitstreams of coded picture is worth being developed.
|出版物ステータス||Published - 2001 12 1|
|イベント||Proceedings of the 22nd Picture Coding Symposium: PCS-2001 - Seoul, Korea, Republic of|
継続期間: 2001 4 25 → 2001 4 27
|Conference||Proceedings of the 22nd Picture Coding Symposium: PCS-2001|
|国||Korea, Republic of|
|期間||01/4/25 → 01/4/27|
ASJC Scopus subject areas