最小互信息量用于单输入单输出的盲解卷问题

Minimum Mutual Information (MMI) Method for single-Input Single-Output (SISO) Blind Deconvolution

本文利用最小互信息量标准处理盲解卷问题．首先，分别通过Edgeworth拓展和Gram—Charlier拓展方法近似输出的概率密度函数．继而估计互信息量．最后利用最小化互信息量原理求得优化的解卷滤波器系数，仿真实验证明了本文算法的有效性．

This paper presents a new algorithm for blind deconvolution based on minimum mutual informationcriterion. Edgeworth expansion and Gram-Cbarlier expansion are used to approximate the possibility density function(p.d.f) of the output. Then the mutual information is estimated from which the optimal coefficients ofdeconvolution filter is calculated under the criterion of muinimum mutual information.simulation experiental resultsshowed the validity of the algorithm.