Variable-p Iteratively Weighted Algorithm for Image Reconstruction

摘要

In this work, a novel variable-p iteratively weighted algorithm is proposed for image reconstruction. The proposed algorithm introduced a dynamically adjusting strategy for the variable p in the nonconvex l_p(0< 1) norm optimization problem, which can be solved via an iteratively reweighted algorithm. Thus, the image reconstruction procedure could start with the l_1 norm problem which is easy to implement, and approach the original l_0 norm problem during the iteration. Numerical experiments indicate that the proposed algorithm can reconstruct the images with fewer measurements than existing fixed-P algorithms.