A Hybrid Orthogonal Forward-Backward Pursuit Algorithm for Partial Fourier Multiple Measurement Vectors Problem

Algorithm 4

HMP.

Input: The measurement matrix , the

measurement vectors where , the

sparsity .

Initialization: Let the residual vectors

where and the support set .

Iteration:

(1) Let the temporary support

set

where is the recovered signal

by applying OMP on the residual vector.

(2) Let

(3) The residual vector is updated as

where .

(4) If , let ,

and go back to step (1); otherwise,

terminate the iteration and let the final support set

.

Output: Compute the estimated signals by

(5).