基于小波变换模极大值的去噪方法研究

小波去噪在信号处理中得到广泛应用。目前常用的方法主要有Donoho提出的阈值法和Mallat提出的模极大值法。模极大值法是一种经典的小波去噪方法,噪声的模极大值的幅度随尺度的增大而迅速减小,而正常信号随尺度的增大而增大,因此利用合适尺度的小波变换,容易把噪声从正常信号中剔除。通过试验说明这种方法对白噪声和脉冲噪声都有很好的去噪效果,并与阈值去噪比较,对于高斯白噪声,信噪比比较低的信号,模极大值去噪要优于阈值法去噪;对于脉冲噪声,脉冲噪声点数较多时,模极大值去噪要优于阈值法去噪。

Research in de-noising method based on wavelet transform modulus maxima

Wavelet de-noising is widely used in signal processing. The current methods commonly used are mainly the threshold method proposed by Donoho and the modulus maxima proposed by Mallat. The modulus maxima method is a classical wavelet de-noising method. The modulus maxima amplitude of the noise decreases rapidly with the increase of scale, while the normal signals amplitude increases with the increase of scale. So appropriate scale of wavelet transform can easily remove noise from the normal signal. Test proves the method has very good denoising effect on white noise and impulse noise. For Gaussian white noise, when SNR is low, there are many impulse points, modulus maxima denoising method is superior to threshold de-noising; for impulse noise, when the modulus maxima de-noising is superior to threshold de-noising.