Invention Grant
US07656320B2 Difference coding adaptive context model using counting 有权
差分编码自适应上下文模型使用计数

Difference coding adaptive context model using counting
Abstract:
Techniques for improving encoding and decoding data are described herein. According to one embodiment, it is determined whether a current context can encode a retrieved symbol. The current context includes a plurality entries, each representing an encoded symbol, including a count value representing a frequency of the entry being used. A code is generated to a code stream, where the code represents a difference between an index of an entry in the current context associated with the retrieved symbol and a previous index used for encoding a previous symbol, if the current context can encode the retrieved symbol. A count value corresponding to the entry associated with the retrieved symbol is incremented in the current context. The current context is sorted based on count values of all entries in the current context, where the code stream and the literal stream are to be compressed and encoded by a compressor.
Public/Granted literature
Information query
Patent Agency Ranking
0/0