基于对称小波变换的多层系数乘积去噪算法

De-noising algorithm in multi-scale products based on symmetry wavelet transform

为了有效地去除信号中的噪声,提出了一种基于对称小波多层系数乘积的信号去噪算法。该算法利用对称小波能够反映信号突变位置的特点,把相邻的两层细节系数相乘得到模极大值,再加以阈值化和归一化,然后与低层的细节系数相乘,得到降噪的细节系数,重构后得到去噪信号。给出了应用该算法的具体步骤,并且通过仿真实验证明了该算法的有效性。

To remove the noise in the signal effectively, a new signal de-noising algorithm based on multi-scale products in wavelet domain is presented. This algorithm utilizes the symmetrical wavelet characteristic which can refleet the signal protrusive position to obtain the modulus maximum value through multiplying neighboring two detail coefficients each other. The value is limited and normalized, then multiplies with the underlying layer detail coefficient to obtain a new detail coefficient, and restructuring to obtain the de-noising signal. The concrete steps to apply the algorithm are given, and the algorithm validity is proved by simulation experiment.