基于动态规划算法的最优Context量化器设计

Optimized Context Quantizer Design Based on Dynamic Programming Algorithm

提出一种针对多进制信源的最优Context量化器设计方法.该方法不仅综合考虑了量化前后条件概率分布的相似性,同时又将条件位符号的取值相关性作为量化合并的依据,从而使得量化后的Context模型能够最大限度地利用信源间相关性,然后动态规划算法被应用于合并相似的条件概率分布,从而实现Context量化.最后量化器被用于图像的小波压缩编码应用.实验结果表明,量化器能够获得与其他优化量化器相近甚至更好的压缩效果.

The optimal Context quantizer design method aiming at multi-system information sources was put forward.This method not only focused on the similarity of conditional probability distribution before and after quantization,but at the same time,takes the value correlation of conditional source symbols as the gist of quantization combination so as to use greatly information source correlation of Con-text model after quantization.Then the dynamic programming algorithm was used to merge similar conditional probability distribution to realize Context quantization.Last,the quantizer was used into image wavelet compression coding application.The experiment results show that the quantizer can obtain the similar or even better compression effect compared with others.