Joint Multichannel Deconvolution and Blind Source Separation
Published in Signal Processing with Adaptive Sparse Structured Representations (SPARS) workshop, 2017
Abstract
In the real world, current Blind Source Separation (BSS) methods are limited since extra instrumental effects like blurring have not been taken into account. Therefore, a more rigorous BSS has to be solved jointly with a deconvolution problem, yielding a new inverse problem: deconvolution BSS (DBSS). We introduce an innovative DBSS approach, called DecGMCA, which is based on sparse signal modeling and an efficient alternative projected least square algorithm. Numerical results demonstrate the performance of DecGMCA and highlight the advantage of jointly solving BSS and deconvolution instead of considering these two problems independently.