Title :
A direct solution for blind separation of sources
Author :
Mansour, A. ; Jutten, C.
Author_Institution :
Inst. Nat. Polytech. de Grenoble, France
fDate :
3/1/1996 12:00:00 AM
Abstract :
Proposes a direct estimation of the mixing matrix in the source separation problem. As a new result, we prove that for non-Gaussian sources, the mixing matrix can be obtained by solving a fourth-degree polynomial equation that derives from nonlinear equations only involving fourth-order cumulants
Keywords :
estimation theory; higher order statistics; nonlinear equations; polynomial matrices; signal processing; blind separation; direct estimation; direct solution; fourth-degree polynomial equation; fourth-order cumulants; mixing matrix; nonGaussian sources; nonlinear equations; source separation; Higher order statistics; Minimization methods; Nonlinear equations; Nonlinear systems; Polynomials; Signal processing; Source separation;
Journal_Title :
Signal Processing, IEEE Transactions on