基于Givens矩阵和联合非线性不相关的盲源分离新算法  

Novel Blind Source Separation Algorithm Based on Givens Matrix and Joint Non-linear Uncorrelatedness

作　　者：赵礼翔[1] 刘国庆[2] 

机构地区：[1]南京工业大学电子与信息工程学院,南京211816 [2]南京工业大学理学院,南京211816

出　　处：《计算机科学》2015年第5期149-152,共4页Computer Science

基　　金：江苏省自然科学基金(BK2011238);南京气象雷达开放实验室研究基金(BJG201103)资助

摘　　要：针对源信号统计独立的盲源分离(Blind Source Separation,BSS)问题,提出了一种基于Givens矩阵和联合非线性不相关的盲源分离新算法。由于分离信号独立性的度量是影响算法有效性的重要因素,因此首先提出了一种改进的度量独立性的方法,该方法以独立源信号的联合非线性不相关来度量独立性;其次,结合Givens矩阵可以对分离矩阵施加正交性约束且能减少要估计参数个数的性质,将盲源分离问题转化成无约束优化问题,并利用拟牛顿法中的BFGS算法求解该无约束优化问题,得到分离矩阵;最后,通过模拟混合信号和真实语音混合信号的分离实验验证了该算法的有效性。A novel algorithm based on Givens matrix and joint non-linear uncorrelatedness for blind source separation （BSS） where sources are statistically independent was proposed. Firstly, we measured independence with the joint nonlinear uncorrelatedness of independent sources, since the measurement is the key to the effectiveness of the BSS algorithm. Secondly,based on the property that Givens matrices can impose the orthogonal constraint on the separation matrix and meanwhile decrease the number of parameters to be estimated by half approximately, we converted the BSS problem into an unconstrained optimization problem which is then solved by BFGS algorithm of quasi-Newton method. Finally,the separation for simulated mixed signals and real mixed voice signals shows the effectiveness of our algorithrn.

关 键 词：盲源分离 独立成分分析 非线性不相关 Givens矩阵 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]