Block Inverse-free Sparse Bayesian Learning for Block Sparse Signal Recovery

Compressed sensing has important applications in many areas and there are many approaches for sparse signals recovery. Sparse Bayesian learning is a popular recovery method. Recently an inverse-free sparse Bayesian learning (IFSBL) has been proposed, which has low computational complexity without matrix inverse. In practice, the non-zero elements of the sparse signal tend to have certain structural characteristics and block sparse recovery algorithms are required. The main work of this paper is to extend the IFSBL algorithm to the case of block sparse signals and propose a block IFSBL algorithm, which utilizes the cluster-structured signal prior model for sparse signal recovery. The simulation results show that it can effectively reconstruct arbitrary block sparse signals with fast computational speed.