We investigated the quality of the spectral estimation of incandescent and fluorescent illuminants using Non-negative Matrix Factorization (NMF), Independent Component Analysis (ICA) and a direct pseudo-inverse approach. We simulated the response of a commercial digital CCD camera coupled or not with coloured filters to a set of natural and artificial illuminants. None of the recovery algorithms used here needed information about spectral sensitivities of the camera sensors or eigenvectors to estimate the spectral power distributions of illuminants. Although nonnegative algorithms can reduce the computational cost of spectral devices, experiments show that ICA and direct pseudo-inverse methods consistently outperforms the NMF approaches, even for fluorescent lights and even using a reduced training set of illuminants.