Blind primary colorant spectral separation combining ICA and POCS non-negative matrix factorization

Traditional printing technologies use metamerism to match color between original objects and reproduced images under certain lighting constraints. These techniques become unsatisfactory for applications demanding high color accuracy where lighting conditions are subject to change, for example, artwork reproduction. Therefore, it is desirable to reproduce the image using colorants whose spectrum is closest to the original print. To estimate the original object primary colorant spectrum, we propose a blind spectral separation algorithm by combining independent component analysis (ICA) and non-negative matrix factorization (NNMF). Our experimental results show a satisfactory match between the original and the estimated primaries´ spectrums.