| 摘要 |
PURPOSE: To decide a symbol probability in accordance with the context of the symbol and also substantially reduce a defect being incidental to a conventional arithmetic encoder by evaluating a sub-pattern to decide a prediction possible position PP and a prediction impossible position UP, using the probability of the respective symbols concerning the respective symbols in PP and applying arithmetic encoding through the use of a combination probability which can be appeared concerning the respective symbols in UP. CONSTITUTION: The context of the symbols is evaluated and the probability with a condition based on the context is decided. Thus, the symbols are classified into the prediction impossible point UP where the probability is equel to 1/n and the prediction possible point PP. The encoding of the prediction possible point PP is executed by a standard arithmetic encoding method based on the symbol probability. When the prediction impossible point UP is processed, one of the two methods is selected in accordance with which is requested, loss(fuzzy) encoding or non-loss encoding.
|
