冗長性のある離散ウェーブレット変換のサブバンド適応ディジタルフィルタへの適用に関する研究

概要

論文の詳細を見る
Adaptive digital filtering is a widespread technique in many applications. In some applications such as acoustic echo cancellation or active noise control, very large adaptive digital filters are used in a system identification context. In addition, subband techniques for adaptive digital filters have been proposed to improve their convergence. On the other hand, discrete wavelet transform is one of the methods of signal processing, and multiresolusion analysis is the well-known method of them. This multiresolusion analysis forms filter banks and can be used for the subband filters. However, it has a demerit of lack of shift invariance, and the adaptive digital filter using it has little advantage of convergence. Therefore, in this paper, redundant discrete wavelet transform is proposed to be used for the subband adaptive digital filter. Thereby, many kinds of orthogonal wavelets, which have been prepared for the discrete wavelet transform, can be used for the widespread adaptive digital filters. In addition, some simulations are given to show the effectiveness of the proposed method in the system identification context, moreover, acoustic echo cancellation with speech signals is simulated.
社団法人日本機械学会の論文
2007-10-25