摘要
基于多尺度分析,引入Mallat提出的补偿因子削减非连续二进尺度对小波系数造成的影响;利用信号和噪声奇异指数不同的特点判断噪声模极大值,并对其所在选通域进行直线平滑处理;把阈值去噪和Witkin的尺度跟踪理论结合,有效恢复了1尺度细节系数。仿真实验结果表明,这种改进后的方法能保护信号的原有信息,有着更好的去噪效果且计算量较小。
Based on multiscale analysis, it is introduced the compensation factors given by Mallat to wavelet coefficients of each scale to reduce the influence of scale discretization. The algorithm determines the modulus maxima caused by noise according to the Lipschitz, and then processes the area between its adjacent modulus minima. What’s more, the method of threshold denoising is combined with scale trading theory by Witkin to recover the coefficients of 1-scale efficiently. Simulation result shows this improved method has a better effect on denoising, can protect the original information, and needs fewer computations.
出处
《系统仿真学报》
CAS
CSCD
北大核心
2005年第4期838-840,843,共4页
Journal of System Simulation
关键词
模极大值
奇异指数
小波去噪
信号重建
仿真
modulus maxima
lipschitz
wavelet denoising
signal reconstruction
simulation