The Weighted L2,1 Minimization for Partially Known Support

A weighted L2,1 minimization is proposed for signal reconstruction from a limited number of measurements when partial support information is known. The reconstruction error bound of the weighted L2,1 minimization is obtained and our sufficient condition is shown to be better than δ3K<13\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\delta _{3K}<{\frac{1}{\sqrt{3}}}$$\end{document} if the estimated support is at least 50 % accurate. Experiments are given for larynx image sequence to illustrate the validity of the proposed method.