结合小波零树的分形图像编码  

Fractal image coding combined with wavelet subtree

作　　者：王端芳[1] 姜威[1] 

机构地区：[1]山东大学信息科学与工程学院,山东济南250100

出　　处：《系统工程与电子技术》2005年第6期1120-1122,共3页Systems Engineering and Electronics

摘　　要：图像经多级小波变换以后系数有两大特点:一是图像能量主要集中在低频带,即在小波系数的高频子带部分会出现大量的零树;二是同方向多级小波子带之间具有相似性。根据小波系数的这两大特点,再结合人眼的特性是对高频失真不敏感,特别是对对角线方向的高频失真最不敏感,而对低频失真特别敏感的特性,提出结合小波零树的分形图像编码方法。理论分析和实验结果表明,在保证图像质量的情况下,能够获得较高的压缩比。The wavelet coefficients of an image present two characteristics when the image is processed by wavelets: one is that the power of an image is strongly concentrated in low-resolution region, in other words, there are many zerotrees in the wavelet subimage; the other is that there is a similarity between wavelet subimages. According to the above two characteristics of wavelet coefficients, and based on the fact that human eyes are sensitive to low resolution distortion but not sensitive to high resolution distortion, especially not sensitive to that in diagonal directions. An image compression algorithm based on wavelet subtree using fractal prediction coding and zerotree coding is proposed. Theoretical analysis and experimental results presented show that a high compression ratio is achieved when the quality of image is not declined.

关 键 词：零树编码 分形编码 人类视觉特性 

分 类 号：TN919.81[电子电信—通信与信息系统]