| |
| Recently, many applications about the recovery of block sparse signals have arisen, which can be casted as the recovery of a block sparse signal $x$ from a measurement equation $y=Phi x $. Here, $Phi$ is a known measurement matrix, and assumed to be a block-concatenation of Toeplitz matrices. In this paper, StOMP is extended to the block sparse case, and an algorithm tBlock-StOMP is proposed. Specifically, tBlock-StOMP combines advantages of StOMP with the structural characteristics of $x$ and $Phi$ to pursue high efficiency in block sparse signal recovery. Furthermore, a modification to the tBlock-StOMP is proposed, termed mtBlock-StOMP. Compared with many other recovery algorithms, numerical simulations demonstrate that tBlock-StOMP as well as mtBlock-StOMP results in evident effectiveness in block sparse reconstruction problems. |
|
| Keywords:Parameter estimation, compressive sensing, block sparse |
|
