一种基于小波变换的分形零树混合编码方法

A Wavelet Image Compression Algorithm Based on Fractal Coding and Zerotree

为了降低分形编码的复杂度和提高编码效率 ,提出了一种基于小波变换的分形零树混合图象编码方法 .该方法首先利用小波变换将图象分解为不同方向不同分辨率的子图象 ,然后分别在不同分辨率层将这些子图象以类似于零树的结构 ,构成一棵棵的小波子树 (wavelet subtree) ,最后在编码时 ,对每一棵小波子树进行具体的分析 ,或采用分形编码 (fractal coding) ,或采用零树编码 (zerotree coding) .与传统的基于小波变换的分形编码相比 ,此算法在充分利用子带图象间的相似性和块内存在的自相似性的同时 ,也充分利用了小波变换后子图象块内 ,特别是高分辨层的子图象块内所存在的大量局域性的冗余性 .实验结果表明 ,此方法在较大的压缩范围内 ,都能够获得较好的压缩效果 .

In order to achieve a high image compression ratio in fractal cloding, the ability of fractal coding to predict wavelet coefficients is anyalyzed and the traditional way of fractal coding is found to be not able to effectively predict the entire wavelet coefficients and leads to a not very good coding result. A hybrid image compression algorithm based on wavelet transforming using fractal coding and zerotree coding that can make up for this flaw effectively is presented in this paper. First, the image is discomposed into a series of subimages in different orientations and different resolutions by wavelet transform, then the subimages in the same orientations but different resolutions are formed into wavelet subtrees, just like zerotree,at last ,the wavelet subtrees are coded by the way of either fractal or zerotree coding according to the size of error when coding.. This algorithm made a effective use of the redundance within subimages as well as the self-similarities within subimages and the similarities cross scales compared with traditional fractal image coding based on wavelet transforming. The experimental with this algorithm presented in this paper also show that the proposed algorithm can obtain a good compression result in a broad compression rate scale.