摘要
为了克服小波阈值去噪中硬阈值小波系数不连续和软阈值估计小波系数与分解小波系数之间恒定偏差的缺点,改进的阈值去噪方法被相继提出。文章根据高斯白噪声和信号在小波变换以后得到的小波系数呈现不同的特性,基于噪声方差提出一种新算法。最后通过MATLAB仿真验证该算法在信噪比、均方根误差、相关系数、信噪比增益4个去噪指标的效果。
In order to overcome the disadvantages of wavelet-coefficient discontinuity in hard threshold and constant deviation between estimation of wavelet-coefficient and decomposition of wavelet-coefficient in soft threshold,many improved threshold de-noising schemes without the two disadvantages has been proposed. In this paper,we have achieved a novel algorithm based on noise variance by analyzing different characteristics of wavelet coefficient between Gauss white noise and signal after wavelet transform. The experimental results by MATLAB simulation show that the proposed algorithm has four de-noising indexes advantages in SNR,RMSE,correlation coefficients and SNR gain.
出处
《微型机与应用》
2016年第21期58-60,共3页
Microcomputer & Its Applications
关键词
小波阈值
小波变换
噪声方差
去噪指标
wavelet threshold
wavelet transform
noise variance
de-noising index
