具有一定消失矩的优化线性相位滤波器  

Optimized linear-phase filterbanks with given degree of vanishing moments

作　　者：龚劬[1] 

机构地区：[1]重庆大学数理学院信息与计算科学系,重庆400044

出　　处：《高校应用数学学报（A辑）》2004年第4期479-485,共7页Applied Mathematics A Journal of Chinese Universities(Ser.A)

摘　　要：利用消失矩特性和编码误差最小,给出了一类有限长线性相位的双正交小波滤波器组BNVF的构造方法,BNVF的综合低通和分解高通的系数为二进分数,将其用于图像的分解与重构时,可避免一些乘法运算.三组BNVF用于图像变换编码时,其压缩性能不亚于Cohen,Daubechies等构造的9/7步滤波器组CDF-9/7,且计算复杂度更低.The paper proposes a new class of linear phase biorthogonal wavelet filterbanks called BNVF,whose synthesis 'low-pass' filter has dyadic fraction coefficients, which is an attractive feature in realizing multiplication-free discrete wavelet transformation. The method constructing BNVF includes two steps. First is to determine the synthesis 'low-pass' filter coefficients with minimal length based on the perfect reconstruction condition, given the same degree of vanishing moments for synthetic scaling function and wavelet pair. Second is to determine the analysis 'low-pass' filter coefficients to minimize the error of the reconstructed image after coding and decoding,subject to the perfect reconstruction condition and linear phase condition. Three filterbanks in this family are verified to have competitive compression potential and much smaller computational complexity to the 9/7 tap filterbanks CDF-9/7 by Cohen,Daubechies, et al.

关 键 词：图像压缩 双正交小波 最优化 小波变换 变换编码 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]