摘要
在相关去噪和模极大值去噪的基础上,提出了一种基于小波窗口相关的模极大值去噪算法。即先用小波窗口相关法对最大尺度的小波系数进行预处理,再用模极大值法去除各层系数的噪声。该算法不仅克服了通常相关去噪算法中小波系数对偏移敏感的缺点,避免了阈值选择受噪声影响的问题,同时,它也解决了模极大值算法中由于小尺度上噪声影响较大而造成的对信号小波系数定位不准的问题,减少了模极大值法的累积误差。仿真实验验证了新方法的有效性,特别是信噪比较低时,该方法的效果尤为显著。
On the basis of correlation denoising and max-modular denoising, a denoising arithmetic based on the max-modular of wavelet windows is proposed which uses the wavelet window to deal with the largest scale of the wavelet coefficients to filter some noise coefficients first, then wipes off noises with the method of max-modular. Different from some conventional correlation denoising methods, the wavelet coefficients of this algorithm are not sensitive to the excursion, and it avoids the threshold selection problem affected by noise seriously. Meanwhile, it can solve the problem of orientation inaccuracy to wavelet coefficients due to the greater impact of noises on small scale and reduces the cumulate errors of max-modular. The experiment testifies the validity of the new method, especially in low SNR.
出处
《系统工程与电子技术》
EI
CSCD
北大核心
2008年第10期1844-1846,1913,共4页
Systems Engineering and Electronics
关键词
尺度相关性
模极大值
去噪
小波系数
二相编码信号
scalerelativity
max-modular
denoising
wavelet coefficient, two-phase code signal
