An Improved Multiwavelet Denoising Method Using Neighboring Coefficients
Bing S Huang · Jisuanji fangzhen · 2008
Wavelet analysis has been an important theory of signal denoising. Multiwavelet possesses such very important properties in signal processing as orthogonality, symmetry, and short support, thus making up for the shortcomings of scalar wavelet, and taking on broader application prospects. When a signal is decomposed by mutiwavelet, there will be correlation not only between neighboring coefficients but also cross the scales. This paper first uses multiwavelet or translation-invariant multiwavelet to decompose the signal polluted by white noise, then takes advantage of the cross-scale correlations to control some large coefficients of noise on small scales while using neighboring coefficients method to determine the threshold. An improved multiwavelet denoising method using neighboring coefficients is proposed in this paper. The experimental results show that the proposed method outperforms those traditional methods such as vector threshold and neighboring coefficients.