An Effective Signal De-noising Algorithm Combining Optimal Wavelet Packet Basis and Translation-Invariant Algorithm

Abstract

An effective algorithm combining optimal wavelet packet basis (OWPB) algorithm and translation-invariant (TI) wavelet transform algorithm was put forward. Based on wavelet packet analysis, according to least cost function principle, OWPB of signal decomposition was obtained, which would better match the requested localization in the frequency domain and ensure an efficient de-noising: Firstly, the estimated OWPB coefficients were gained by using thresholding methods, and then based on inverse discrete wavelet packet transform (IDWPT), the de-noised signals were obtained by using estimated OWPB coefficients. Because wavelet thresholding de-noising methods would result in Pseudo-Gibbs phenomenon in the neighborhood of discontinuities, in order to suppress it, this study adopt TI wavelet transform algorithm. The results of experiment indicated the algorithm proposed in the pager was better than traditional wavelet packet transform (WPT) thresholding de-noising methods.