摘要 |
Conditioning states are selected for coding a source from a plurality of potential conditioning states. The potential conditioning states represent a series of context events, such as a plurality of binary context events. According to a rule, potential conditioning states are partitioned into groups, such as by partitioning the potential conditioning states according to the number or significant context events in each of the conditioning states. The conditioning states within each of the groups are then collected into layers, and then the conditioning states in the layer are quantized to form a next layer of collective conditioning states. Coding efficiency indicators, such as context quantization efficiencies, are calculated for each of the layers. Based on the context quantization efficiencies, layers in each of the groups are selected for use in a context model for coding the source.
|