一种新的小波阈值去噪方法研究

A New Wavelet Threshold De-noising Method

本文研究用组合小波对含噪信号的去噪问题,对均方根误差、信噪比、平滑度等信号指标进行多指标融合作为去噪效果的评价参数,根据评价参数确定最佳小波分解尺度、小波去噪最佳小波基和最佳阈值。新的去噪方法与传统的去噪方法相比,克服了硬阈值不连续的缺点,还克服了软阈值中估计小波系数与分解小波系数之间的恒定偏差的缺点。MATLAB仿真结果表明,新的阈值函数的去噪效果在各指标上都优于传统的阈值去噪方法。

In this paper, the problem of eliminating noise from signals that contains noise is researclaea by using combined wavelet. It merges root-mean-square error, signal to noise ratio, smoothness, and other signal indicators, serving as evaluation parameters for the effect of eliminating noise. Based on the evaluation parame- ters, the best scale of wavelet decomposition, the best wavelet basis of eliminating noise form wavelet and the best optimal threshold are determined. Compared with the traditional methods of eliminating noise, the new ones overcome the shortcoming of the hard threshold, which is discontinuous. The new methods also overcome the constant deviation between estimated wavelet coefficients and the exploded wavelet coefficients in the soft threshold. MATLAB simulation results show that the effect of eliminating noise by new threshold is better than traditional ones in all indicators.