一种基于显著不相关检验的近距分形图象编码方法

A Neighboring Search Fratcal lmage Coding Based on Testing Significant Uncorrelation of Blocks

提出一种基于显著不相关检验的近距分形图象编码方法。对于标准测试图象（Ｌｅｎａ２５６×２５６×８ｐｐｂ），这种方法与子块分类方法相比，以解码图象质量（ＰＳＮＲ）下降２～３（ｄＢ）为代价，编码速度提高了７０～２００倍，且压缩比还有一定的提高；与普通近距分形方法相比，在解码图象质量（ＰＳＮＲ）不下降的情况下，编码速度提高约３２％；若以解码图象质量（ＰＳＮＲ）下降０．２～１（ｄＢ）为代价，编码速度提高３～９倍。

In this paper,we proposed a very fast fractal image coding method, which is based on neighboring search for the matching of image blocks and testing of significant uncorrelation of blocks.For the standard test image(Lena 256×256×8 ppb),comparing with the classification method, we recorded acceleration factors from 70 up to 200 with 2(dB) to 3(dB) degradation of PSNR and some improvement of the compression ratio; comparing with the ordinary neighboring search method, we recorded a 32% acceleration factor without any degradation of PSNR, and if 0.2(dB) to 1(dB) degradation of PSNR are permitted, acceleration factors from 3 to 9 are recorded.