基于Fisher分类和自适应阈值的分形图像编码方法

Fractal Image Coding Based on Fisher Classification and Adaptive Threshold

关键词:  分形图像编码  自适应阈值  四叉树分割  Fisher分类

Key words:  fractal image coding  adaptive threshold  quadtree partition  Fisher classification

摘要:

根据分割的迭代函数系统(PIFS)理论,在Fisher分类方法的基础上,依据分类后图像块的特征使用自适应阈值搜索与range块属于同一类的相匹配的domain块来进行分形图像编码,并对典型测试图像进行实验,与传统的固定阈值方法的性能进行比较。结果,在相同比特率上,相对于传统固定阈值方法,自适应阈值方法极大地提高了压缩比,减少了编码的时间,自适应阈值方法得到的解码图像质量明显好于传统的固定阈值方法。

Abstract:

Based on the theory of Partitioned Iterated Function Systems (PIFS, for short)and Fisher's classification, this paper proposes a new adaptive method in fractal image compression.The method changes the MSE threshold linearly, to search the best-match domain block in the same class as range block.Finally, the proposed method enhences largely the compression ratio and reduces the encoding time at the same bit rate, as compared with that of the constant threshold method.Meanwhile, the perceptual quality of the decoded image is better than that of the latter.