基于自适应扫描的遥感图像压缩算法

Remote sensing image compression based on adaptive context quantization

目的为提高基于上下文算术编码器的效率。方法首先分析了上下文模型以及上下文量化对信源熵的影响,得到了上下文量化的两个优化准则,并根据准则建立了一种时变的上下文性能模型。结果给出了一种自适应位平面扫描顺序,改善了上下文条件概率的时变稳定性,提高了编码过程中对上下文条件概率的估计精度。结论该方法较顺序扫描方式在同码率下有一定的PSNR提高,具有比顺序扫描更优的编码效率。

Aim In order to improve coding effciency of context based arithmetic coder.Methods Source entropy change under influence of context model and context quantization is analyzed,then two criteria of optimum context quantization and time-variant context performance mode were established.Results Based on that,an adaptive scan method is proposed to improve the stability of time-variance.Conclusion Higher PSNR at same bpp in experiment of remote sensing image compression shows that it has better performance than normal scan method.