ウェーブレット変換を用いた適応線スペクトル強調器の提案

概要

論文の詳細を見る
Adaptive line enhancer which is one of the application of adaptive digital filter can extract only periodic signals buried in a broad-band noise, under the condition of the frequencies of the periodic signals are unknown. However it is known that adaptive line enhancer gets impossible to remove noise for a certain period after a strong pulse noise is put to the adaptive line enhancer with input signal. In this paper, an adaptive.line enhancer by using wavelet transform is proposed to get over the problem. Here, discrete wavelet transform is used for the proposed method, however, the discrete wavelet transform based on multiresolution analysis has a demerit of the lack of shift invariance inevitably. Hence considering the demerit of discrete wavelet transform, in this paper, redundant discrete wavelet transform is used which realizes the shift invariance. The effectiveness of the proposed method is shown through simulations. Moreover voice signals that are sampled from human voice are used in simulation to remove noise to show the effectiveness of the proposed method even for voice signals.
社団法人日本機械学会の論文
2006-10-25