基于Gabor特征分解的高斯混合非线性滤波算法  

Gauss Hybrid Nonlinear Filter Design Based on Gabor Feature Decomposition

作　　者：高菲菲[1] 

机构地区：[1]内蒙古财经大学统计与数学学院,内蒙古呼和浩特010070

出　　处：《科技通报》2015年第12期88-90,共3页Bulletin of Science and Technology

摘　　要：传统的窄带信号检测滤波器采用IIR自适应线谱增强滤波算法,对信号特征分解的阶数要求高,导致非线性失真,提出一种基于Gabor特征分解的高斯混合非线性滤波器设计算法,在IIR滤波器设计的基础上,对信号进行尺度和时延估计,构建自适应高阶累积量滤波设计方法,采用高阶累积量对窄带信号进行均方一致估计,对Gabor特征函数Taylor级数展开,求得高斯混合非线性滤波器的带宽参数,最后实现高斯混合非线性滤波器设计改进,提高对窄带信号的检测性能。仿真结果表明,该算法具有较好的滤波性能,可以明显地抑制色噪声的影响,提高信号增益达到20 d B。Narrow band signal detection filter is used in the traditional IIR adaptive line enhancement algorithm, order de-composition on signal feature requirements, resulting in nonlinear distortion, this paper puts forward a Gabor feature decom-position algorithm based on Gauss mixture nonlinear filter design, based on IIR filter design, scale and time delay estima-tion of signal, to construct an adaptive high order cumulants filter design method, using high order cumulant of mean square consistent estimation of narrowband signals, the characteristic function expansion on the Gabor Taylor series, the band-width parameter obtained Gauss mixed nonlinear filter, finally realize the Gauss improvement of mixed nonlinear filter de-sign, improve the detection performance of the narrowband signal. The simulation results show that the proposed algorithm has good filtering performance and can obviously suppress the color noise and improve the signal gain of 20 dB.

关 键 词：滤波器 非线性 特征分解 自适应 

分 类 号：TN911.7[电子电信—通信与信息系统]