Abstract: Traditional fractal image coding is very time consuming, so its application is limited. To speed up the encoding speed, this paper proposes fast fractal coding algorithm based on semi-cross trace feature. The algorithm includes: defining semi-cross trace feature, achieving the relationship between the root mean square of sub-blocks error and the semi-cross trace feature of sub-blocks. Experimental results show that: when decoded image PSNR increases of about average 0.63 dB, the encoding speeds up average 55 times; the algorithm improves coding performance and improves the quality of decoded image in case of unchanged encoding time, compared with main diagonal sum feature from the literature[10] and cross trace feature from the literature[11]; the fast encoding algorithm based on sub-block feature, the coding performance is related to the complexity of image. The more abundance image details, the worse coding performance.

Key words: fractal, fractal image coding, coding performance, sub-block feature, semi-cross trace

摘要： 传统的分形图像编码时间过长，限制了它的应用。为了加快编码速度，提出基于半叉迹特征的快速分形编码算法，该算法主要包括：定义子块半叉迹特征，导出子块的均方根误差与子块半叉迹特征之间的关系。实验结果表明：该算法较基本分析编码算法，在解码图像PSNR平均提高约0.63 dB的情况下，平均加快编码速度55倍；较文献[10]提出的主对角和算法和文献[11]提出的叉迹算法，在编码时间不变的情况下，改善了编码性能，提高了解码图像质量；基于子块特征的快速编码算法，其编码性能与图像的复杂程度有关，细节信息越丰富的图像，编码性能越差。

关键词: 分形, 分形图像编码, 编码性能, 子块特征, 半叉迹