基于参数设计字典的稀疏表示方法

The Method of Sparse Representation Based on Parameter Dictionary Design

稀疏化是压缩感知理论的关键,对信号恰当的稀疏表示能提高恢复的精度.本文提出一种基于参数字典的稀疏表示方法,把参数字典设计作为一个优化问题来分析,通过交替迭代的方式求得参数方程的可行解,进而生成参数字典.本文的参数字典设计方法较其他方法而言能获得较优的近似解,且该方法产生的优化字典更符合紧框架特性.

Sparsity is the key to compressed sensing theory. The appropriate sparse representation can im- prove the accuracy for signal recovering. In this article, we propose a method based on the parameter dictionary design for sparse representation. This means that the parameter dictionary design is considered as an optimization problem. Further, through the alternating iterative method, we can obtain a feasible solution of parametric equations and produce a parameter dictionary. The parameter dictionary design can acquire the approximate optimum solution compared to other methods. In the parameter dictionary design