Blind source separation using time-frequency information matrix given by several wavelet transforms

被引：0

作者：

Ashino, Ryuichi ^{[1
]}

Fujita, Keiko ^{[2
]}

Mandai, Takeshi ^{[3
]}

Morimoto, Akira ^{[1
]}

Nishihara, Kiyoaki ^{[1
]}

机构：

[1] Osaka Kyoiku Univ, Osaka 5828582, Japan

[2] Saga Univ, Saga 8408502, Japan

[3] Osaka Electrocommun Univ, Res Ctr Phys & Math, Osaka 5728530, Japan

来源：

INFORMATION-AN INTERNATIONAL INTERDISCIPLINARY JOURNAL | 2007年 / 10卷 / 05期

关键词：

blind source separation; time-frequency information; wavelet transform;

D O I：

暂无

中图分类号：

T [工业技术];

学科分类号：

08 ;

摘要：

The blind source separation method for spatial mixing problem is discussed. A matrix called time-frequency information matrix is composed of continuous wavelet transforms of observed signals with respect to several different wavelet functions. The proposed method detects the number of sources and separates sources using the time-frequency information matrix. Algorithms axe given, and numerical experiments demonstrate the proposed method works well.

引用

页码：555 / 568

页数：14

共 9 条

[1]

Balan R., 2000, P ICA 2000 19 22 JUN

[2]

Cichocki A., 2003, Adaptive Blind Signal and Image Processing

[3]

Fujita K., 2005, EA200512 IEICE, P37

[4]

FUJITA K, IN PRESS APPL MATH C

[5]

JOURJINE A, 2000, IEEE INT C AC SPEECH

[6] CHARACTERIZATION OF SIGNALS FROM MULTISCALE EDGES [J].

MALLAT, S ;

ZHONG, S .

IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE, 1992, 14 (07) :710-732

[7]

MORIMOTO A, 2005, EA200512 IEICE, P31

[8]

MORIMOTO A, 2006, P 4 INT C INF 4 IR C, P164

[9]

NAPOLETANI D, 200247 U MAR TR

← 1 →