非平稳分形随机信号波形估计的最优门限方法

Waveform Estimation of Non-Stationary Fractal Stochastic Signals Using Optimum Threshold Technique

本文用基于最小均方误差准则的最优门限方法估计叠加高斯白噪声的分形布朗运动，并给出 其离散小波变换分解级数确定方法.与多尺度维纳滤波相比，本方法不需估计1/f类分形信号 的方差，且其离散小波变换分解级数可预先确定，因此有着更好的实用性和可操作性.

Based on the minimum mean-square error, this paper offers an optimum threshold method of estimating the 1/f-type fractal signals with the additive white noise. The paper also provides the method of determining the wavelet decomposition scale. Compared with the multiscale Wiener filter, the optimum threshold method is more practical and maneuverable because it doesn't estimate the variance of 1/f-type fractal signals and the wavelet decomposition scale can be determined in advance.