| This paper mainly includes three parts:the first two chapters are the first part,which introduce the split minimization problem,specifically the split minimization problem of difference of l1-L2 norms.In addition,we introduce the difference of convex algorithm(DCA)and the of Ming-fukushima algorithm,and use these two algorithms to solve this problem.The second part is the third chapter.In this paper,we propose a more general normal term of convex difference,which is denoted as l1-Lp,where 1<p<? We prove the relationship between them and signal sparsity,and then use DCA and Mingˇfukushima algorithm to solve the split minimization problem with such penalty terms,and give the proof of the convergence respectively The third part is the fourth chapter.In this chapter,we apply the model of this kind of problem to reconstruct the sparse signal.Numerical experiments show that when we choose an proper p,l1-Lp outperforms than the difference of l1-L2 norms in motivate sparseness. |
PDF Full Text Request