An algebraic solution to independent component analysis
- 作者：Yamaguchi ; Taro ; Itoh ; Kazuyoshi
- 关键词：Independent component analysis ; Blind separation ; Signal processing ; Algebra ; Image processing ; 07.05.P
- 刊名：Optics Communications
- 出版年：2000
- 期刊代码：50_00304018
- 类别：ph
- 出版时间：May 1, 2000
- 卷：178
- 期：1-3
- 页码：59-64
- 文件大小：260 K
摘要
The independent component analysis (ICA) is the technique of extracting original signals from many independent sources without a priori information on the sources and the mixing process. We proposed a novel algebraic solution to ICA. Independent components are recovered by solving simultaneous equations derived from the definition of the independence. The proposed algorithm is evaluated by an experiment where original images are extracted from mixed images. The reduction of the processing time is achieved with the comparable levels of accuracy to conventional algorithms.