一种JPEG2000算术编码器的优化算法与实现

通过对索引表和概率间隔区间更新条件的分析,提出了一种JPEG2000算术编码器的部分并行优化算法.在连续编码两个数据对时,通过预测间隔区间的变化,可以一次完成索引表和间隔区间的更新,从而减小了编码数据的关联性,实现了算术编码器的部分并行编码.设计了基于3级流水线的JPEG2000算术编码器,并通过了FPGA验证.试验结果表明,该算法平均每个时钟编码1.58对数据,比每个时钟编码1对数据的普通算法,编码效率提高了58%.

An Optimized Algorithm and its Implementation for Arithmetic Encoder in JPEG2000

After the analysis of the updating condition for index table and probability interval,an optimized algorithm based on partial parallel scheme was presented for arithmetic encoder(AE) in JPEG2000.Based on the prediction of probability interval,the index table and the probability interval can be updated at one time when coding two contexts consequently.So the contexts can be coded in a partial parallel scheme by minifying the relationship between the coded data.A 3stage pipelined MQ encoder was proposed,which was simulated and verified on FPGA.Experiment results show that the proposed algorithm,processing 1.58 data-pairs per clock in average,can achieve 58% improvement encoder compared with the normal algorithm which processes one data-pair per clock.