基于谱分解技术的两通道因果稳定IIR滤波器组的设计(英文)

提出一种新的两通道因果稳定IIR滤波器组的设计方法.首先基于所定义的一种余弦滚降IIR滤波器,两通道滤波器组所要求的平坦条件完全被满足;其次该方法是依赖于分析滤波器的功率谱特性,而不是该滤波器的幅度频率特性来设计滤波器组.IIR滤波器的设计问题可转化成一个准凸问题,采用凸优化算法来求解;然后对所得优化解进行谱分解,得到目标滤波器组.该滤波器组具有合理的完全重构误差,并且其多相因子的分母相同,简化了完全重构条件.采用该滤波器组作为约束非线性优化法的有效初值,进一步优化得到完全重构2通道IIR滤波器组.     

线性迭代算法设计双正交余弦调制滤波器组

介绍了一种用于设计双正交余弦调制滤波器组的线性化方法。该方法中，目标函数和完全重构条件均被表示成为原型滤波器系数矢量的二次函数形式。使用拉格朗日乘数方法，算法通过迭代求解线性方程来获得期望的原型滤波器系数矢量。该方法最大优点在于设计过程简单易行并且计算效率非常高。设计实例表明，该算法对于双正交余弦调制滤波器组的设计是非常有效的。     

线性迭代算法设计双正交余弦调制滤波器组

介绍了一种用于设计双正交余弦调制滤波器组的线性化方法。该方法中，目标函数和完全重构条件均被表示成为原型滤波器系数矢量的二次函数形式。使用拉格朗日乘数方法，算法通过迭代求解线性方程来获得期望的原型滤波器系数矢量。该方法最大优点在于设计过程简单易行并且计算效率非常高。设计实例表明,该算法对于双正交余弦调制滤波器组的设计是非常有效的。     

基于线性神经网络的余弦调制滤波器组设计

通过拟合满足准确重建条件的理想频率响应训练网络,提取训练完成的网络权值和阈值,构造原型滤波器,经过余弦调制得到满足要求的余弦调制滤波器组.仿真表明原型滤波器阶数较少,近似准确重建余弦调制滤波器组.     

余弦调制滤波器组的设计研究

介绍了一种近似准确重建(NPR)的余弦调制滤波器组(CMFB)的设计方法,与传统设计相比,该设计方法是集高效性与良好的原型滤波器设计于一体.利用线性迭代方程的方法来解决阻带的优化问题,迭代误差小,迭代收敛快,线性度好,阻带衰耗特性也明显优于其它传统的设计方法.     

完全重构的非均匀滤波器组设计

讨论了完全重构的非均匀多通道滤波器组的设计方法,分析了多种条件下的设计问题,针对不同的条件,建立了相应的优化双目标函数,将进化策略和连续二次规划算法相结合,提出了一种混合优化设计方法,可解双目标函数问题,具有收敛性好,计算代价适中等特点.设计结果表明,用混合优化算法可以得到性能良好的非均匀滤波器组.     

基于余弦调制构造M带紧支撑双正交小波

作为经典两带小波的拓展,基于完全重构 (PR) 滤波器理论研究了任意M带小波的性质,提出了基于余弦调制的提升结构构造M带紧支撑双正交小波的新算法.与通常的非线性优化方法相比,所提出的算法需要优化的参数更少,稳定性和鲁棒性更好,且可以用离散余弦变换(DCT)的快速算法实现.数值实验结果表明,这种方法不但实现简单快捷,而且所得到的多带小波性能比现有方法有显著改进.     

基于余弦调制滤波器组的多载波调制

首先介绍了Ｍ带滤波器组的概念,然后给出基于此概念的ＣＭＦＢ余弦调制滤波器组的设计及实现,并根据计算机仿真结果,与ＤＭＴ在性能和复杂度上作了比较。     

离散余弦列率滤波器的设计及应用

为在信号滤波领域发挥离散余弦变换（ＤＣＴ）潜在的优良特性,提出了离散余弦列率滤波的时域卷积算法,引入一种新型的线性相位数字滤波器－离散余弦率滤波器（ＤＣＳＦ）导出了任意长度一维和二维ＤＣＳＦ的设计公式,讨论了这种滤波器在广义标量维纳滤波和倍频程于带滤波中的应用,实验结果表明,ＤＣＳＦ可实现无形波失真的良好滤波效果,ＤＣＳＦ滤波算法简单,其计算量（乘法和加法）仅为采用快速ＤＣＴ（ＦＣＴ）的算法的１／     

近似重构正交镜像滤波器组的设计

为设计一种正交镜像滤波器组(QMF)的原型滤波器,使用高边瓣衰减(SLFOR)的可变性和组合窗函数设计了低通原型滤波器,并且应用线性优化使近似重构的误差最小.结果表明,该方法程序运行简单,优化算法使计算机处理的速度加快,而且高边瓣衰减(SLFOR)组合窗减少了能量的泄漏,防止了子带间的混叠.     

Analysis of PR and near-PR cosine-modulated filter banks with odd and even numbers of channels

It is well known that M-channel filter banks, satisfying either the perfect-reconstruction (PR) or near-PR property, can be obtained by cosine modulation of a linear-phase prototype filter. In this work, we investigated the effects of choosing odd or even number of channels on the performance of both PR and near-PR cosine-modulated filter banks (CMFBs). Given the same m, a performance degradation happens in the PR case with odd numbered M. Contrarily, the choice of even or odd M has little effects on the whole system quality for near-PR one. In addition, we made some comparisons between the constrained optimization and the Parks-McClellan algorithm with cosine roll-off characteristic in designing near-PR CMFBs. Detailed analyses and numerical comparisons show that the PR and near-PR systems have their individual characteristics in terms of M and m selection. Studies here can provide useful and practical guidelines for choosing right filter systems.     

Design of nonuniform cosine-modulated filter-banks with the perfect-reconstruction property and arbitrary filter lengths

In this paper, a method is proposed to construct recombination nonuniform cosine-modulated filter banks (CMFBs) with a perfect-reconstruction property and with arbitrary filter lengths. In this filter bank (FB) structure, certain channels of an original M-channel uniform FB are merged by a set of transmultiplexers (TMUXs), yielding nonuniform outputs, and the length restrictions on the original CMFB and the recombination TMUXs are avoided, allowing them to have arbitrary filter lengths. By imposing certain improved matching conditions on the prototype filters, nonuniform FBs with arbitrary filter lengths and with good properties can be obtained. As a result, the flexibility of selecting modulation types and filter lengths is greatly increased in the design of recombination nonuniform CMFBs. Another benefit due to the flexibility is that the system delay can be reduced.     

Design of nonuniform cosine-modulated filter-banks with the perfect-reconstruction property and arbitrary filter lengths

In this paper, a method is proposed to construct recombination nonuniform cosine-modulated filter banks (CMFBs) with a perfect-reconstruction property and with arbitrary filter lengths. In this filter bank (FB) structure, certain channels of an original M-channel uniform FB are merged by a set of transmultiplexers (TMUXs), yielding nonuniform outputs, and the length restrictions on the original CMFB and the recombination TMUXs are avoided, allowing them to have arbitrary filter lengths. By imposing certain improved matching conditions on the prototype filters, nonuniform FBs with arbitrary filter lengths and with good properties can be obtained. As a result, the flexibility of selecting modulation types and filter lengths is greatly increased in the design of recombination nonuniform CMFBs. Another benefit due to the flexibility is that the system delay can be reduced.     

两通道完全重构全相位FIR滤波器组的设计

提出了两通道完全重构全相位FIR滤波器组的设计算法，理论证明了无窗和双窗情况下全相位滤波的3个重要性质．并针对不同的滤波器长度N和频率向量的设置（传统对称或偶对称），设计出了4种全相位半带滤波器．再对半带滤波器进行谱分解，构造出了各分析和综合滤波器．理论和仿真实验表明，全相位PR滤波器组具有较高的重构精度，滤波器长度N为8时的重构误差比CDF9／7双正交小波滤波器组低3个数量级．     

数字信道化器中高阶精确重构滤波器组设计方法与量化分析

数字信道化技术广泛应用于新一代卫星通信系统的星上处理平台中,而精确重构滤波器组的设计是其中的关键技术。针对现有滤波器设计方法通道数较低,难以满足高分辨率数字信道化器需求的问题,提出了一种基于频率取样法和最小均方误差逼近的高阶精确重构原型滤波器的设计方法。该方法首先利用频率取样法求解一个近似精确重构原型滤波器;然后在最小均方误差准则下,用一个由两通道无损网格建模的滤波器对该近似精确重构原型滤波器进行最优拟合。数值实验表明,与Parks-Mc Clellan(PM)算法相比,该方法能设计得到更高通道数的滤波器组;并且原理简单,设计复杂度很低。滤波器性能的量化分析结果表明,综合考虑多相分量长度与量化位宽的设计思想,能够平衡实现复杂度与性能指标,对于实际系统具有重要的参考价值。     

线性相位二维递归数字滤波器组的设计

提出一种线性相位二维递归数字滤波器组的设计方法.这种设计是以 S 域一维(1-D)贝塞尔函数为原型的低通滤波器的传递函数,通过一维 S 域至二维 Z 域的双线性变换,实现线性相位二维递归数字滤波器.所实现的二维数字滤波器不仅保持贝塞尔滤波器通带内相位线性、阻带内相位近似线性的优点,而且能利用一维至二维的双线性变换压缩过渡带,使过渡带变得陡峭、阻带衰减增加.采用本文的设计方法,利用低阶的线性相位递归数字滤波器组,即可实现高性能的二维分析与综合数字滤波器组,完成图象的子带分割与重建功能.计算机仿真实验证明所提出的设计方法是正确的、有效的.     

用于滤波器组多载波系统的新型滤波器优化设计算法

为了提高滤波器多载波系统的设计性能,针对多载波系统的原型函数优化问题,提出了一种启发式优化算法—雕塑算法。通过分析原型滤波器的性能评价指标,建立原型滤波器优化设计的数学模型,将原型滤波器的优化设计归结为一个多变量联合求取最小值的问题,并采用提出的雕塑算法进行优化求解。以基于模拟退火算法的滤波器优化设计算法作为比较对象,评估了雕塑算法在算法有效性及求解质量等方面的性能。仿真结果表明,雕塑算法设计的滤波器在阻带衰减幅度、过渡带宽度、时域拖尾衰减速率等方面表现优异,在应用于经典的滤波多音调制系统中,系统误码率也较低。     

Design of perfect reconstruction rational sampling filter banks

The design of rational sampling filter banks based on a recombination structure can be formulated as a problem with two objective functions to be optimized. A new hybrid optimization method for designing perfectreconstruction rational sampling filter banks is presented, which can be used to solve a class of problems with two objective functions. This method is of good convergence and mezzo calculation cost. Satisfactory results free of aliasing in analysis and synthesis filters can be obtained by the proposed method.