基于人眼视觉特性加权的率失真优化编码算法

本文提出了一种基于人眼视觉特性加权的率失真优化的嵌入式静态图像编码算法。率失真优化是按照率失真斜率递减的顺序编码系数位,最先编码的位具有最大的率失真斜率,每编码一位都会使失真减少最大。率失真斜率的计算仅仅是利用MQ自适应算术编码器的概率状态估计表而进行的查表过程,率失真门限方法的编码速度比搜索最大的率失真斜率更快。基于人眼视觉特性的加权因子改变编码位序而不是编码值,该算法有更快的编码速度以及较好的压缩效果。

A Rate-Distortion Optimized Coding Algorithm Based on Human Visual Weighting

An embedded still image coding algorithm with rate-distortion optimization based on human visual weighting is proposed in this paper. Rate-distortion optimized embedding code coefficient with decreasing R-D slope,so the first coding bits are ones mat have the steepest R-D slope. The distortion can be reduced to the maximum extent per coding bit. The calculation of R-D slope is just a lookup table operation with the help of probability estimation table of MQ-adaptive arithmetic coder. The coding speed of R-D threshold is much faster than the search for the maximum R-D slope. The human visual weighting factor change the order of coding rather than the value of coding. The algorithm has much faster coding speed and better compression effect.