论文部分内容阅读
针对基于分类的快速分形编码方法存在着编码速度与解码质量间的矛盾,鉴于Krawtchouk矩不变量具有在仿射变换下保持不变的特性和核模糊聚类在处理非线性问题上的突出优势,本文首次将这两者引入到分形编码中,提出了基于Krawtchouk矩不变量和核模糊聚类的自适应分类快速分形编码方法。首先根据Domain块的方差将其粗分类,再根据Domain块的Krawtchouk矩不变量利用核模糊聚类对Domain块细分类。实验结果表明,与其他基于分类的快速分形编码方法相比,在解码图像质量提高的同时,大大加快了分形编码的速度。
The fast fractal coding method based on classification has the contradiction between encoding speed and decoding quality. In view of Krawtchouk moment invariant has the property of being invariant under affine transformation and the outstanding advantage of kernel fuzzy clustering in dealing with nonlinear problem, This paper introduces the two into fractal coding for the first time, and proposes an adaptive fast classification fractal coding method based on Krawtchouk moment invariants and kernel-based fuzzy clustering. Firstly, we classify the Domain block according to the variance of the Domain block, then classify the Domain block by kernel fuzzy clustering according to the Krawtchouk moment invariants of the Domain block. Experimental results show that compared with other fast fractal coding methods based on classification, the speed of fractal coding is greatly accelerated while the quality of decoded image is improved.