基于自相似系数聚类的快速分形编码研究

Clustering method of fractal image coding based on self-similarity

自相似是分形理论中重要的特征,广泛地应用于分形图像编码。聚类分析是将数据对象分组成为多个类,在同一个类中的对象之间具有较高的相似度。通过对图像自相关模型的研究,提出了判断图像自相似性强弱的方法,定义了自相似系数来表征图像的自相似程度,并将自相似系数与聚类算法相结合,改进了分形图像编码,实验结果表明,该方法在不影响重建图像质量的情况下,加快了分形编码的速度。

Self-similarity is the important characters of fractal theory, and applied to the fractal image coding widely. Clustering method means grouped all the data together, data in the same class has higher similarity. A method that estimates image＇ s self-similarity is presented and the self-similarity coefficient of image＇ s self-similarity is defined through studying the self-similarity model of image. As combining the self-similarity coefficient and clustering algorithm, the fractal encoding of image is developed. Experimental results indicate that this method accelerates fractal image encoding, and not reduces the quality of reconstruct image.