基于小波变换的分形与零树混合图像压缩算法

Hybrid image coding algorithm based on wavelet zerotrees and fractal coding

提出了一种基于小波变换的分形与零树混合的图像压缩方法.该算法采用小波变换将图像分解为不同方向、不同分辨率的子图像,在不同分辨率层将这些子图像以类似于零树的结构构成一棵棵的小波子树;在对每一棵小波子树编码时,根据最小误差标准进行分形编码,或零树编码.本算法不但充分利用了子带图像间的相似性和块内的自相似性,而且充分利用了小波变换后子图像块内(特别是高分辨层的子图像块内)存在的大量的局域性的冗余性.实验结果表明,与传统的基于小波变换的分形编码相比,该方法在较大的压缩范围内,都能够获得好的压缩结果.

The hybrid image coding algorithm presented here is based on wavelet zerotrees and fractal coding,At first,the image is decomposed into a series of subimages in different orientations and at different resolutions by wavelet transform,and then the subimages in the same orientations but with different resolutions are formed into wavelet subtrees,like zerotree; finally,the wavelet subtrees are coded by means of fractal or zerotree coding. This algorithm effectively reduces the redundancies within subimages as well as the self-similarities within subimages;and the similarities cross scales are compared with traditional fractal image coding based on wavelet transforming.Experiments show that the proposed algorithm has good performance within a broad compression rate scale.