计算机科学 ›› 2015, Vol. 42 ›› Issue (5): 149-152.doi: 10.11896/j.issn.1002-137X.2015.05.029

• 网络与通信 • 上一篇    下一篇

基于Givens矩阵和联合非线性不相关的盲源分离新算法

赵礼翔,刘国庆   

  1. 南京工业大学电子与信息工程学院 南京211816,南京工业大学理学院 南京211816
  • 出版日期:2018-11-14 发布日期:2018-11-14
  • 基金资助:
    本文受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

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

ZHAO Li-xiang and LIU Guo-qing   

  • Online:2018-11-14 Published:2018-11-14

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。

关键词: 盲源分离,独立成分分析,非线性不相关,Givens矩阵

Abstract: 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 non-linear 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 algorithm.

Key words: Blind source separation,Independent component analysis,Non-linear uncorrelatedness,Givens matrix

[1] Meganem I,Deville Y,Hosseini S,et al.Linear-Quadratic Blind Source Separation Using NMF to Unmix Urban Hyperspectral Images [J].IEEE Transactions on Signal Processing,2014,62(7):1822-1833
[2] Mauldin F R,Lin D,Hossack J A.The singular value filter:a general filter design strategy for PCA-based signal separation in medical ultrasound imaging [J].IEEE Transactions on Medical Imaging,2011,30(11):1951-1964
[3] Wang L,Ding H,Yin F.A region-growing permutation alignment approach in frequency-domain blind source separation of speech mixtures [J].IEEE Transactions on Audio,Speech,and Language Processing,2011,19(3):549-557
[4] Xie S,Yang L,Yang J M,et al.Time-frequency approach to underdetermined blind source separation[J].IEEE Transactions on Neural Networks and Learning Systems,2012,23(2):306-316
[5] Kleinsteuber M,Shen H.Blind source separation with compressively sensed linear mixtures[J].Signal Processing Letters,IEEE,2012,19(2):107-110
[6] Hyvarinen A.Fast and robust fixed-point algorithms for independent component analysis[J].IEEE Transactions on Neural Networks,1999,10(3):626-634
[7] Hyvrinen A.Independent component analysis:recent advanced [J].Physical and Engineering Sciences,Philosophical Transactions of the Royal Society A:Mathematical,2013,371(1984):411-430
[8] 茆诗松,程依明,濮晓龙.概率论与数理统计教程[M].北京:高等教育出版社,2004:165-234
[9] 朱孝龙,保铮,张贤达.基于分阶段学习的盲信号分离[J].中国科学:E辑,2002,32(5):693-703
[10] 李思怡,王永威,黄琰,等.一种基于自然梯度的两步盲源分离算法[J].微电子学与计算机,2013,30(006):169-172
[11] Watkins D S.Fundamentals of Matrix Computations [M].Second Edition.New York:John Wiley & Sons,2002:192-194
[12] Degroot M H,Schervish M J.Probability and Statistics(4th ed)[M].Boston:Addison Wesley,2011:140-140
[13] Hyvarinen A.Independence Component Analysis [M].NewYork:John Wiley & Sons,2001:342-344
[14] 解可新,韩健,林友联.最优化方法(修订版)[M].天津:天津大学出版社,2004:114-123
[15] Hyvarinen A.Fast and robust fixed-point algorithms for independent component analysis[J].IEEE Transactions on Neural Networks,1999,10(3):626-634
[16] Cardoso J F,Souloumiac A.Blind beamforming for non-Gaussian signals[C]∥IEE Proceedings F.IET Digital Library,1993,140(6):362-370
[17] Bach F R,Jordan M I.Kernel independent component analysis [J].The Journal of Machine Learning Research,2003,3:1-48
[18] Yang H H,Amari S.Adaptive online learning algorithms forblind separation:maximum entropy and minimum mutual information[J].Neural computation,1997,9(7):1457-1482

No related articles found!
Viewed
Full text


Abstract

Cited

  Shared   
  Discussed   
No Suggested Reading articles found!