【摘要】 目前分形图像压缩的最主要问题是其编码时间太长,这主要是因为在分形编码时,对每一个待编码值域块都需要比较数量巨大的定义域块才能找到它的最优匹配块。通过深入分析分形编码过程,文章首先提出了一种图像的矩不变量,它在灰度仿射变换下保持不变,并以此作为图像块的特征来为分形编码中的图像块进行分类,从而得到了一种基于矩不变量的快速分形编码方法:将定义域块按其矩不变量进行分类,在编码时对每一个待编码值域块,其最优匹配块只在其同类或相邻类的定义域块中寻找,从而大大地减少了定义域块的比较数目,缩短了编码时间。实验证明,与已有的分类方法相比,该文方法在解码图像质量基本满意的基础上,极大地提高了分形编码的速度。更多还原

【Abstract】 The nowadays fractal image compression schemes suffers from long encoding time ,because of considerable comparisons with domain blocks for each range block to find its best-match domain blocks.This article first proposes a kind of quadrature of digital images,which is invariant under grey-scale affine maps,then it can be used as a feature of image blocks to classify them,hence a novel and fast method for fractal image encoding is proposed:It can first classifies all the domain blocks into several categories based on their quadratures that has proposed above;then in the procedure of encoding,for every range block to be encoded,compute its quadrature and search for its best -match domain block in those with the same category or adjacent categories,hence reducing dramatically the number of domain blocks needed to be compared with.Comparing with other existed methods,the experimental results demonstrate its efficiency on speed-up of fractal image compression,with only little degeneration of decoded images.更多还原