Non-negative matrix factorisation for blind source separation in wavelet transform domain

Abstract

This paper describes a new multilevel decomposition method for the separation of convolutive image mixtures. The proposed method uses an adaptive quincunx lifting scheme (AQLS) based on wavelet decomposition to preprocess the input data, followed by a non-negative matrix factorisation whose role is to unmix the decomposed images. The unmixed images are, thereafter, reconstructed using the inverse of AQLS transform. Experiments carried out on images from various origins showed that the proposed method yields better results than many widely used blind source separation algorithms.