Multiresolution Subband Blind Source Separation: Models and Methods

Abstract

The goal of this paper is to review the multiresolution subband Blind Source Separation (MRSBSS) methods, classify them into different categories, and identify new trends. After the perspective of MRSBSS is briefly presented, the general and detailed definition of the MSBSS model is given, the relationships among BSS, Independent Component Analysis (ICA) and MRSBSS method are discussed simultaneously. Then, the state-of-theart MRSBSS algorithms: adaptive filter based method and sparsity based multiscale ICA method are constructed and overviewed in detail from different theory base. The paper concludes by discussing the impact of multiresolution subband on BSS research and outlining potential future research directions and applications.