期刊名称：International Journal on Electrical Engineering and Informatics
印刷版ISSN：2085-6830
出版年度：2018
卷号：10
期号：1
页码：37-50
DOI：10.15676/ijeei.2018.10.1.3
出版社：School of Electrical Engineering and Informatics
摘要：At the moment, there are two main methods of solving the compressive sensing(CS) reconstruction problem which are the convex optimization and the greedy algorithm.Convex optimization has good reconstruction stability but very slow in computation.Greedy algorithm, on the other hand, is very fast but less stable. A fast and stable CSreconstruction algorithm is necessary for a better provision of CS in practical application.In this paper we proposed a CS reconstruction algorithm using L1-norm minimization viaL2-norm minimization. This method is based on geometrical interpretation of L1-normminimization of the reconstruction problem and the fact that the Euclidean distancebetween L1-norm and L2-norm solution lie closely. In other word, if L2-norm solution isfound, then direction to the L1-norm solution is on the shortest path connecting them. Thisapproach offers a simpler computation. Computer simulation showed that proposedalgorithm has better stability than the greedy algorithm and faster computation than theconvex optimization. The proposed algorithm thus provides an alternative solution for CSreconstruction problem when a balance between speed and stability is required.
