基于回归分析的嵌入式块编码预测截断算法

Prediction truncation based on regression analysis for JPEG2000 embedded block coding

针对JPEG2000中基于率失真优化后压缩处理的嵌入式块编码算法所存在的运行复杂、浪费大量编码时间等问题,提出了一种有效提高编码效率的方法,即基于回归分析的嵌入式块编码预测截断算法。该算法根据多元线性回归分析得到的回归方程来预测截断点,避免了对将被截断的位平面进行扫描编码,并舍弃了复杂耗时的率失真计算与搜索,有效地加快了嵌入式块编码算法的运行速度。实验结果表明,算法获得了可观的编码速度提升收益。

To solve the problems of execution complexity and time waste in compression process induced by Post Compression Rate Distortion（PCRD） Optimization in embedded block coding in JPEG2000,this paper proposes an effective method to promote coding efficiency-embedded block coding with prediction truncation based on regression analysis.This algorithm predicts the truncation points based on regression formula derived from linear regression analysis.It avoids scanning and encoding the bit-planes that can be truncated,and discards the time-consuming rate-distortion truncation processing,so that the algorithm is able to speed up the compression process.The experiment results show the considerable speed promotion.