基于提升框架的整数小波变换  

Wavelet Transforms that Map Integers to Integers Based on Lifting Scheme

作　　者：吴永宏[1] 潘泉[1] 张洪才[1] 张绍武[1] 张云龙[1] 

机构地区：[1]西北工业大学自动控制系,西安710072

出　　处：《电子与信息学报》2004年第4期659-663,共5页Journal of Electronics & Information Technology

基　　金：教育部"跨世纪优秀人才培养计划项目"基金;国家自然科学基金(60172037)资助课题

摘　　要：提升框架可以从一个新的角度来构造小波滤波器,它的核心是确定预测算子和更新算子。该文 引进了多项式局部拟合方法来求取预测算子和更新算子,对每步提升下取整,就实现了整数小波变换.仿真 结果表明,该方法应用于二维数字图像,在一定失真的前提下,具有较高的压缩比,而且小波系数与尺度系 数都是整数,不需要量化步骤.因此,它对于整数数据压缩具有重要的意义.The lifting scheme is a new idea of constructing wavelet filters. Its key element is to find the predict operator and the update operator. The polynomial local regression is used to find these operators in the paper. These lifting steps take integers downwardly. Then the wavelet transforms that map integers to integers are realized. The algorithm is applied to two dimensions digital image processing, and the simulation indicates that compression ratio is high with allowable distortion. The scale coefficients and wavelet coefficients are integers, thus the quantification is not needed. This brings the significan importance of the proposed algorithm for integer data compression.

关 键 词：提升框架 整数小波变换 预测算子 更新算子 

分 类 号：TN911.72[电子电信—通信与信息系统]