压缩感知中信号重构的极大熵方法.pdfVIP

应用戴学 MATHEMA TlCA APPLlCATA 2010 ， 23(2):345钢352 Maximum Entropy Function Method for Signal Reconstruction in Compressed Sensing WANG Tian-jing( 王天荆)1 ， YANG Zhen( 杨震户，WU Ye-jun( 吴业军)3 (1. Collegeol Scíence ,Nanjíng Universíty of Technology ,Nanjing 210009.China;2.1nsti- tute 01 Signαl and lnformation Processing. Nanjing University of Pυsts Telecommuni- cations .Nanjing 210003 .China ;3. Deρartment 01 Bωsic Cour5e •N anj ing 1nstitute 01 Tech- nology ， Nωnjing 211167 .China) Abstract: 丁he emerging theory of CompressOO Sensing (CS) has 100 to the remarkable results that compressible signal can be reconstructed using only a small number of mea阳rements. Significant at- tention in CS has been focusOO on Basis Pursuit (BP) .exchanging the sparseness constraint with the II norm. In order to overcome the nonsmooth problem in 1 norm.this paper proposes a new Maxi- 1 mum Entropy Function Method (MEFM) to solve the I 1 optimization problem via smoothing the ob- jective function with maximum entropy function. lntimately relating to homotopy method. MEFM provides a systematic approach for deriving the global optimal sparse solution. Finally. the numerical results show that it is an effective technique for signal reconstruction. ln a CS framework. MEFM is a usefully alternating method to solve the II optimization problem. Key words: Compressed sensing; Nonsmooth optimization I Homotopy method; Maxi- mum entropy function method CLC Number: TN912 AMS(2000)Subject Classification:94A99 Document code: A Artícle ID: 1001-9847(2010) 02-0345-08 1. Introduction The recent evolution of sensing devices has triggered new research in the fields of com- pression and coding. Current compression