小波去噪中噪声估计失真判别方法

研究小波阈值去噪时存在的噪声估计失真问题。当采用常见的阈值确定方法对含有较强高频分量的信号进行小波去噪时,小波分析的频带能量泄漏现象会导致噪声估计失真,从而使小波阈值去噪出现较大的偏差。从小波分解的d1细节层和d2细节层的相关性角度,揭示d2细节层频带能量泄漏对噪声估计影响的规律,提出根据d1、d2细节层的最大相关系数判别噪声估计失真的方法。最后,给出解决噪声估计失真的方法。实验表明,该方法可以很好地判别小波去噪中是否出现噪声估计失真,可以避免出现去噪后信号有用信息损失严重的问题。

Identification Method for Noise Estimation Distortion inWavelet Denoising

The noise estimate distortion in the wavelet threshold denoising is researched. When the signal containing some high frequency components is denoised by using the wavelet denoising with usual threshold selection methods, the frequency band energy leakage of the wavelet analysis results in the increase of the noise estimate of the wavelet denoising and makes an error appear in the process of the wavelet threshold denoising. In view of the relevance between the d1 detail level and the d2 detail level, the rules of the noise estimate affected by the frequency band energy leakage from the d2 detail level are analyzed. The identification method for noise estimate distortion is proposed by using the maximum correlation coefficient of the d1 detail level and the d2 detail level. Finally, the method restraining the noise estimate distortion is proposed. The results show that the noise estimate distortion can be distinguished by using the above method, and the problem of the serious useful information loss in the wavelet denoising process can be avoided.