Recovery of block sparse signals from non-uniform finite wavelet frame transform
Sunder Deep, Khole Timothy Poumai, S. K. Kaushik · Physica Scripta · 2025
Abstract This paper investigates block-structured, non-uniform extensions of orthogonal and threshold greedy algorithms in finite wavelet frame transforms. A sufficient condition is derived under which these algorithms exactly recover a block sparse vector of size N 0. Furthermore, B-spark is introduced as a blockwise generalization of a matrix’s spark. It provides an alternative sufficient condition that guarantees the recovered solution is the sparsest possible. Finally, numerical experiments are presented to evaluate the success rates of the proposed BOGA algorithm on both uniform and non-uniform block sparse signals.