Abstract: Fractal image encoding is a promising lossy compression technique in terms of achievable compression ratios and decoded image quality，but it suffers from long encoding time.Time is spent on searching out the best-matched block to each of range blocks in a usually-large domain pool.This paper proposes an accelerating scheme based on the categories of fractional box-counting dimension，by discussing their relation between the texture feature of image blocks and fractional box-counting dimension.Experiment results on three popular 512×512 test images show that，compared to the baseline fractal algorithm with the full search，the proposed algorithm not only can achieve the speed-up of about 30 times，but also can obtain the better decoded image quality.

Key words: image compression, fractal, fractal image encoding, fractional box-counting dimension

摘要： 分形图像编码是一种很有前途的限失真编码方法。该方法具有复杂度高、计算量大的缺点，因此导致编码时间长。编码时间主要花费于在一个通常较大的码本中搜索每个输入子块的最佳匹配块。针对这个问题，通过分析图像子块的纹理特征与分数盒维数的关系，提出一个基于分数盒维数分类的快速编码算法。实验结果显示：对三幅512×512标准测试图像，与全搜索基本分形编码算法相比，该算法既能加快编码速度30倍左右，也能实现更好的解码图像质量。

关键词: 图像压缩, 分形, 分形图像编码, 分数盒维数