一种基于最大熵准则的盲解卷积改进算法

独立源信号的卷积混合比线性混合更接近真实情况。利用非参数概率密度估计方法———Parzen窗函数估计法 ,提出了一种具有良好适应性的盲解卷积改进算法。该算法可以在无需知道信号分布形式的情况下 ,较准确地估计出密度函数值 ,且比传统的最大算法中采用固定的概率密度函数估计更接近信号点的真实概率密度。同时 ,此算法还具有无论对规则、单峰分布还是不规则、多峰分布都可以取得较好的估计的优点。因此 ,在理论上 ,改进算法可以获得比传统算法更优越的分离性能且能广泛地应用于具有各种分布的信号。实验结果证实 ,这一算法能有效地从各种分布的信号包括真实语音、图像等构成的卷积混合信号中恢复出原始信号。与最大熵算法相比 ,改进算法具有更好的分离性和更广泛的适用性。

Improved blind deconvolution algorithm using Parzen window estimation

Blind separation of independent sources from their convolution mixtures is a prolem in many real world multi-sensor applications. Based on the information maximization principle, we present a improved blind deconvolution algorithm for mixed signals with any probability density distribution. The algorithm uses probability density estimated by Parzen window instead of by fixed nonlinear function, which is more approximate to the real PDF of sources. With feedback network, algorithm derives the adaptation rules for separate filters and restores the source signals from convolution mixture. Simulation results show that the improved algorithm can extract independent sources from mixed signals with any distribution successfully. Compared with maximization entropy (ME) algorithm, the proposed algorithm has better performance, more robust capability and wider range of application.