The use of ICA in multiplicative noise

Authors:
D. Blanco;B. Mulgrew;S. McLaughlin;D. P. Ruiz;M. C. Carrion
Affiliations:
Department Física Aplicada, Universidad de Granada, Spain;Institute for Digital Communication, University of Edinburgh, UK;Institute for Digital Communication, University of Edinburgh, UK;Department Física Aplicada, Universidad de Granada, Spain;Department Física Aplicada, Universidad de Granada, Spain
Venue:
Neurocomputing
Year:
2006

Citing 4
Cited 3

Independent component analysis, a new concept?

Signal Processing - Special issue on higher order statistics
A fast fixed-point algorithm for independent component analysis

Neural Computation
Statistical characterisation and modelling of SAR images

Signal Processing
ICA in signals with multiplicative noise

IEEE Transactions on Signal Processing - Part I

Independent component analysis in signals with multiplicative noise using fourth-order statistics

Signal Processing
Neural network based audio signal denoising

ICAIT '08 Proceedings of the 2008 International Conference on Advanced Infocomm Technology
Signal denoising in engineering problems through the minimum gradient method

Neurocomputing

Quantified Score

Hi-index	0.01

Visualization

Abstract

When a linear mixture of independent sources is contaminated by multiplicative noise, the problems of blind source separation and feature extraction are highly complex. Specifically, the approach followed by the independent component analysis does not produce proper results. This is because the output of a linear transformation of the noisy data cannot be independent. However, the statistic of this output possesses a special structure that can be used to obtain the original mixture. In this paper, this statistical structure is studied and a general approach to solving the problem is stated, studying how the strategies followed by the standard ICA methods can be adapted in this case. To illustrate the analysis, the results of two different methods that follow the general approach are presented, where the improvement with respect the standard ICA method is clear.