一种混合快速分形图像编码算法

提出了一种采用两步筛选的混合快速分形编码算法。首先将码本按照矩不变量进行分类,然后寻找给定Range块在所属区间的最好匹配码块,对于匹配误差值大于给定阈值的Range块再进行基于熵值的二次编码。与基于矩不变量的算法比较,该方法在峰值信噪比相同的情况下时间效率提高五倍多,与基于信息熵的算法相比,PSNR值提高近一个分贝。

Hybrid-fast fractal image coding algorithm

This paper proposed a fast fractal image coding algorithm based on two-step selection.Firstly,classified all the domain blocks into several categories based on their moment invariant,and then searched the best-matching domain block for every range block to be encoded in the same category or adjacent categories.Finally,coded secondarily the range blocks which were larger than threshold based on their information entropy.The proposed algorithm achieves the speed-up of over 5 times compared with the moment-feature-based fractal algorithm for the same PSNR.Experimental results show that the proposed hybrid method can improve the speed of fractal coding and the quality of decoded image.