Compared with the other statistical methods, ICA provides a more powerful data representation than PCA because PCA considers the second order moments only and it uncorrelates data, while ICA accounts for higher order statistics and it identifies the independent source components from their linear mixtures [14].