Robust approach for blind separation of noisy mixtures of independent and dependent sources

In this paper, a new Blind Source Separation (BSS) method that handles mixtures of noisy independent / dependent sources is introduced.  We achieve that by  minimizing a criterion that fuses a separating part, based on Kullback–Leibler divergence for either dependent or independent sources, with a regularization part that employs the bilateral total variation (BTV) for the purpose of denoising the observations.  The proposed algorithm utilizes a primal-dual algorithm to remove the noise, while a gradient descent method is implemented to retrieve the signal sources.  Our algorithm has shown its effectiveness and efficiency and also surpassed the standard existing BSS algorithms.

blind source separation
noisy mixtures
dependent sources
bilateral total variation
Kullback–Leibler divergence
  1. Comon P.  Independent component analysis, a new concept?  Signal Processing. 36 (3), 287–314 (1994).
  2. Mansour A., Jutten C.  A direct solution for blind separation of sources.  IEEE Transactions on Signal Processing. 44 (3), 746–748 (1996).
  3. Taleb A., Jutten C.  Entropy optimization.  Artificial Neural Networks – ICANN'97. 529–534 (1997).
  4. Belouchrani A., Abed-Meraim K., Cardoso J.-F., Moulines E.  A blind source separation technique using second-order statistics.  IEEE Transactions on Signal Processing. 45 (2), 434–444 (1997).
  5. Pesquet J.-C., Moreau E.  Cumulant-based independence measures for linear mixtures.  IEEE Trans. Inform. Theory. 47 (5), 1947–1956 (2001).
  6. Cardoso J.-F.  Blind signal separation: statistical principles.  Proceedings of the IEEE. 86 (10), 2009–2025 (1998).
  7. Novey M., Adali T.  ICA by maximization of nongaussianity using complex functions.  2005 IEEE Workshop on Machine Learning for Signal Processing. 21–26 (2005).
  8. Pham D.  Mutual information approach to blind separation of stationary sources.  IEEE Transactions on Information Theory. 48 (7), 1935–1946 (2002).
  9. Keziou A., Fenniri H., Ould Mohamed M., Delaunay G.  Séparations aveugle de sources par minimisation des $\alpha$-divergences.  XXIIe colloque GRETSI, Dijon, 8–11 septembre 2009.
  10. Keziou A., Fenniri H., Ghazdali A., Moreau E.  New blind source separation method of independent/dependent sources.  Signal Processing. 104, 319–324 (2014).
  11. Ghazdali A., Hakim A., Laghrib A., Mamouni N., Raghay S.  A new method for the extraction of fetal ECG from the dependent abdominal signals using blind source separation and adaptive noise cancellation techniques.  Theoretical Biology and Medical Modelling. 12, Article number: 25 (2015).
  12. Mamouni N., Keziou A., Fenniri H., Ghazdali A., Hakim A.  A new convolutive source separation approach for independent/dependent source components.  Digital Signal Processing. 100, 102701 (2020).
  13. Ourdou A., Ghazdali A., Laghrib A., Metrane A.  Blind Separation of Instantaneous Mixtures of Independent/Dependent Sources.  Circuits, Systems, and Signal Processing. 40, 4428–4451 (2021).
  14. Ourdou A., Ghazdali A., Metrane A., Hakim M.  Digital document image restoration using a blind source separation method based on copulas.  In Journal of Physics: Conference Series. 1743, 012034 (2021).
  15. Belouchrani A., Cichocki A.  Robust whitening procedure in blind source separation context.  Electronics letters. 36 (24), 2050–2051 (2000).
  16. Sahmoudi M., Snoussi H., Amin M. G.  Robust approach for blind source separation in non-gaussian noise environments.  Proceedings of ISCCSP, Marrakesh, Morocco, IEEE/EURASIP (2006).
  17. El Rhabi M., Fenniri H., Keziou A., Moreau E.  A robust algorithm for convolutive blind source separation in presence of noise.  Signal Processing. 93 (4), 818–827 (2013).
  18. Ghazdali A., El Rhabi M., Fenniri H., Hakim A., Keziou A.  Blind noisy mixture separation for independent/dependent sources through a regularized criterion on copulas.  Signal Processing. 131, 502–513 (2017).
  19. Tomasi C., Manduchi R.  Bilateral Filtering for Gray and Color Images.  Sixth International Conference on Computer Vision (IEEE Cat. No.98CH36271). 839–846 (1998).
  20. Sklar A.  Fonctions de répartition à n dimensions et leurs marges.  Publ. Inst. Statist. Univ. Paris. 8, 229–231 (1959).
  21. Farsiu S., Robinson D., Elad M., Milanfar P.  Fast and Robust Multi-Frame Super-Resolution.  IEEE Trans. on Image Processing. 13 (10), 1327–1344 (2003).
  22. El Mourabit I., El Rhabi M., Hakim A., Laghrib A., Moreau E.  A new denoising model for multi-frame super-resolution image reconstruction.  Signal Processing. 132, 51–65 (2017).
  23. Afraites L., Hadri A., Laghrib A.  A denoising model adapted for impulse and Gaussian noises using a constrained-PDE.  Inverse Problems. 36 (2), 025006 (2019).
  24. Silverman B. W.  Density estimation for statistics and data analysis.  Monographs on Statistics and Applied Probability, Chapman & Hall, London (1986).
  25. Gumbel E. J. Bivariate exponential distributions.  Journal of the American Statistical Association. 55 (292), 698–707 (1960).
  26. Morgenstern D.  Einfache Beispiele zweidimensionaler Verteilungen.  Mitteilungeblatt für mathematische statistik. 8, 234–235 (1956).
  27. Clayton D. G.  A model for association in bivariate life tables and its application in epidemiological studies of familial tendency in chronic disease incidence.  Biometrika. 65 (1), 141–151 (1978).
  28. Cardoso J. F., Souloumiac A.  Blind beamforming for non-gaussian signals.  IEE Proceedings F (Radar and Signal Processing). 140 (6), 362–370 (1993).
  29. Hyvärinen A., Oja E.  A fast fixed-point algorithm for independent component analysis. Neural Computation. 9 (7), 1483–1492 (1997).
  30. Miller E.-G., Fisher J.-W. III.  Independent components analysis by direct entropy minimization,  Tech. Rep. UCB/CSD-03-1221, University of California at Berkeley, January 2003.