数字图像的盲源分离
|
摘要
独立分量分析(ICA)是近期发展起来的一种非常有效的盲信号处理技术,在许多应用领域正发挥着越来越重要的作用。ICA具有重要的理论和应用价值,在无线通信、声纳、语音处理、图像处理和生物医学等领域具有广泛而诱人的应用前景,在过去的十几年时间里,有关的理论和算法研究都得到了较快的发展,并涌现出了许多有效的算法。目前,ICA已经成为国际上信号处理和人工神经网络等学科领域的一个研究热点。
本论文对ICA算法的基本理论进行了详细的分析,按ICA方法的处理过程划分为三个部分:批处理方法、自适应处理方法及其他的一些处理方法。并研究了ICA方法在图像处理中的应用,包括混合图像的盲分离、图像特征提取与识别、数字图像滤波处理等,并提出了一些有效的改进办法。
论文的主要贡献及创新点包括以下几个方面:
(1)针对日常生活中两幅图像会出现混叠的情况,根据数字化图像灰度值范围有界特性和一般信号盲源分离的方法,提出了一种利用分离比值函数单调特性达到数字图像盲源分离目的的算法。这种算法根据接收端接收到的观察信号,确定观察信号之间比值函数,通过该比值函数的单调性分析,可以找到分离矩阵中的关键值,从而实现图像的盲源分离。该算法无需过多的先验知识,没有统计相关性的约束条件,而且分离速度快,效果明显,鲁棒性很强。
(2)根据独立分量分析方法的特性,针对数字图像滤波提出了一种新颖的方法。该方法将原始图像和整体向右平移一列的图像看作ICA的观测信号,对这两幅强相关性图像组成的矩阵信息进行ICA处理,可以得到一幅锐化图像,表明ICA在这种情况下具有高通滤波的性能。计算机仿真实验结果说明,这种方法提取出的锐化图像具有层次感强,定位精度高等特点,是一种有效的滤波方法。
(3)充分考虑到数字图像信号长度有限和准周期的特性,通过构造一个隐含在分离算法中的预处理模块来提高其时间效率。改进了传统盲源分离算法中预处理方法,利用逐点接收的数字图像信号,实时地调整自适应处理器中网络参数,可以将接收到的混合信号高速、有效地进行分离,同时改进了传统梯度算法的迭代方程,以保证算法的稳定性,可以实现对多个源的混合图像信号进行实时的分离,并且有效提高了分离的精度。该算法复杂度较低,易于工程实现,在电视电话会议、无线通信、雷达成像、生物医学信号分析等领域内将具有广泛的应用前景。
本论文对ICA方法的理论和应用进行了深入的研究,所提出的算法及其在数字图像处理中的应用研究具有一定的创新性,特别是对于ICA在图像处理中的应用研究,具有一定的参考价值和实际意义。
Independent Component Analysis (ICA) is a kind of powerful method for Blind Signal Processing (BSP). It becomes more and more important in widely fields, such as telecommunications, audio signal separation, biomedical signal processing, and image processing. Many literatures on ICA were published and lots of algorithms were proposed during the past ten years in a large number of journals and conference proceedings. ICA becomes one of the most exciting new topics both in the fields of signal processing and artificial neural networks.
In this thesis, the principle and algorithms of ICA are researched in detail. According to ICA algorithms of processing the process is differentiate into three parts: a batch method, adaptive processing methods and other methods of processing. At one time, investigation the ICA method in image processing applications, including blind source separation of mixed images, image feature extraction and recognition, digital image processing such as filtering, and some effective ways to improve it.
The main achievements of this dissertation are put forward:
(1)We present a new method for analytical solution to image blind source separation based on monotone property of separation ratio function aim at the two pictures in the daily life appear to blend, pursuant to the digital image gray value range to have the edge property and common signal blind source separate method. The new algorithms using observation signal of the receiver and obtain ratio function of signal, analysis functional monotonic and find key value of separation matrix, realize image blind source separation. This arithmetic have no use for calculate a prior knowledge, there is no the constraint condition which statistical correlation, and separate speed quickly, effect obvious, the simulation result demonstrated the usefulness of arithmetic.
(2) A novel method of images filtering has been proposed using the independent component analysis of the characteristics. This method will be the original image and the overall translation a row to the right of the image as a signal ICA observation, strong correlation between these two images of the matrix of information for analysis, can be a sharpening images, ICA showed that a high pass filter performance. Computer simulation results also show that this method to extract sharpening image with strong lays of graphics and high location accuracy, is an effective filtering method.
(3) Considering the limited length of the digital image signal and the characteristics of quasi-periodic; we by make a pre-processing module are included in the separation algorithm improve the time efficiency of algorithm. Improvement of the traditional blind source separation algorithm in the pre-processing methods, the use of point by point to receive the digital image signals, real-time adaptive processor to adjust the network parameters can be received by the mixed-signal high-speed and effective separation, while improving the traditional iterative gradient algorithm equation in order to ensure the stability of algorithm can be achieved on a number of mixed-source real-time image signal separation, and effectively improve the separation accuracy.
The thesis investigation ICA of the theory and application, the proposed algorithm and its application in digital image processing of the application possess have innovation , especially for the ICA in image processing applied research, with some reference value and practical significance.
本论文对ICA算法的基本理论进行了详细的分析,按ICA方法的处理过程划分为三个部分:批处理方法、自适应处理方法及其他的一些处理方法。并研究了ICA方法在图像处理中的应用,包括混合图像的盲分离、图像特征提取与识别、数字图像滤波处理等,并提出了一些有效的改进办法。
论文的主要贡献及创新点包括以下几个方面:
(1)针对日常生活中两幅图像会出现混叠的情况,根据数字化图像灰度值范围有界特性和一般信号盲源分离的方法,提出了一种利用分离比值函数单调特性达到数字图像盲源分离目的的算法。这种算法根据接收端接收到的观察信号,确定观察信号之间比值函数,通过该比值函数的单调性分析,可以找到分离矩阵中的关键值,从而实现图像的盲源分离。该算法无需过多的先验知识,没有统计相关性的约束条件,而且分离速度快,效果明显,鲁棒性很强。
(2)根据独立分量分析方法的特性,针对数字图像滤波提出了一种新颖的方法。该方法将原始图像和整体向右平移一列的图像看作ICA的观测信号,对这两幅强相关性图像组成的矩阵信息进行ICA处理,可以得到一幅锐化图像,表明ICA在这种情况下具有高通滤波的性能。计算机仿真实验结果说明,这种方法提取出的锐化图像具有层次感强,定位精度高等特点,是一种有效的滤波方法。
(3)充分考虑到数字图像信号长度有限和准周期的特性,通过构造一个隐含在分离算法中的预处理模块来提高其时间效率。改进了传统盲源分离算法中预处理方法,利用逐点接收的数字图像信号,实时地调整自适应处理器中网络参数,可以将接收到的混合信号高速、有效地进行分离,同时改进了传统梯度算法的迭代方程,以保证算法的稳定性,可以实现对多个源的混合图像信号进行实时的分离,并且有效提高了分离的精度。该算法复杂度较低,易于工程实现,在电视电话会议、无线通信、雷达成像、生物医学信号分析等领域内将具有广泛的应用前景。
本论文对ICA方法的理论和应用进行了深入的研究,所提出的算法及其在数字图像处理中的应用研究具有一定的创新性,特别是对于ICA在图像处理中的应用研究,具有一定的参考价值和实际意义。
Independent Component Analysis (ICA) is a kind of powerful method for Blind Signal Processing (BSP). It becomes more and more important in widely fields, such as telecommunications, audio signal separation, biomedical signal processing, and image processing. Many literatures on ICA were published and lots of algorithms were proposed during the past ten years in a large number of journals and conference proceedings. ICA becomes one of the most exciting new topics both in the fields of signal processing and artificial neural networks.
In this thesis, the principle and algorithms of ICA are researched in detail. According to ICA algorithms of processing the process is differentiate into three parts: a batch method, adaptive processing methods and other methods of processing. At one time, investigation the ICA method in image processing applications, including blind source separation of mixed images, image feature extraction and recognition, digital image processing such as filtering, and some effective ways to improve it.
The main achievements of this dissertation are put forward:
(1)We present a new method for analytical solution to image blind source separation based on monotone property of separation ratio function aim at the two pictures in the daily life appear to blend, pursuant to the digital image gray value range to have the edge property and common signal blind source separate method. The new algorithms using observation signal of the receiver and obtain ratio function of signal, analysis functional monotonic and find key value of separation matrix, realize image blind source separation. This arithmetic have no use for calculate a prior knowledge, there is no the constraint condition which statistical correlation, and separate speed quickly, effect obvious, the simulation result demonstrated the usefulness of arithmetic.
(2) A novel method of images filtering has been proposed using the independent component analysis of the characteristics. This method will be the original image and the overall translation a row to the right of the image as a signal ICA observation, strong correlation between these two images of the matrix of information for analysis, can be a sharpening images, ICA showed that a high pass filter performance. Computer simulation results also show that this method to extract sharpening image with strong lays of graphics and high location accuracy, is an effective filtering method.
(3) Considering the limited length of the digital image signal and the characteristics of quasi-periodic; we by make a pre-processing module are included in the separation algorithm improve the time efficiency of algorithm. Improvement of the traditional blind source separation algorithm in the pre-processing methods, the use of point by point to receive the digital image signals, real-time adaptive processor to adjust the network parameters can be received by the mixed-signal high-speed and effective separation, while improving the traditional iterative gradient algorithm equation in order to ensure the stability of algorithm can be achieved on a number of mixed-source real-time image signal separation, and effectively improve the separation accuracy.
The thesis investigation ICA of the theory and application, the proposed algorithm and its application in digital image processing of the application possess have innovation , especially for the ICA in image processing applied research, with some reference value and practical significance.
引文
[1]张天骐,DS及DS/CDMA信号的估计方法研究,电子科技大学博士论文,2002
[2] Cichocki A, Amari S. Adaptive Blind Signal and Image Processing[M]. New York:Wiley, 2005.
[3] Hyvarinen A, Karhunen J, Oja E. Independent Component Analysis[M].New York:Wiley,2001.
[4] Comon P. Independent Component Analysis, a New Concept?[J].Signal Processing,1994,36(3):287-314.
[5] Moreau E, Maechi O. High-order Contrasts for Self-adaptive Source Separation[J].Int J of Adaptive Control and Signal Processing,1996,10(1):19-46.
[6] Bell A J, Sejnowski T J. An Information-Maximization Approach to Blind Separation and Blind Deconvolution[J], Neural Computation, Vol. 7, No. 6, pp:1129-1159, Nov, 1995
[7] Cardeso J F. Blind Signal Separation : Statistical Principles[J]. Proc IEEE,1998,86(10):2009-2025.
[8] Cardoso J F. Eigen-structure of the fourth-order cumulant tensor with application to the blind source separation problem[J], Proc. IEEE ICASSP, 1990, 15:2655-2658
[9] Kung S Y, Diamantaras K I, Taur J S. Adaptive Principal Component Extraction(APEX) and Applications[J]. IEEE Trans on Signal Processing, 1994,42(5):1202-1217
[10] Jutten C, Herault J. Blind separation of sources, Part 1: An adaptive algorithm based on neuromimetic architecture[J]. Signal Processing,1991,24:1-10
[11] Amari S, Ciehocki A. Adaptive Blind Signal Processing : Neural Network Approaches[J]. Proc IEEE,1998,86(10):2026-2048.
[12] Hyvarinen A. Survey on Independent Component Analysis[J].Neural Computing,1999,2(1):94-128.
[13] Delfosse N,Loubaton P. Adaptive blind separation of independent sources :A deflation approach[J].Signal Processing,1995, 45:59-83.
[14] Cardoso J F, Laheld B H. Equivariant adaptive source separation[J]. IEEE Transactions Signal Processing, 1996,44(12):3017-30
[15] Xi Ren Cao,Ruey Wen Liu.General Approach to Blind Source Separation[J].IEEE Transactions on Signal Processing,Vol 44,No 3,1996
[16] Amari S,Cichocki A, Yang HH. A new learning algorithm for blind signal separation[J], Advances in neural information processing systems, pp: 757-763,1996
[17] Cardoso J F. Infomax and Maximum Likelihood for Source Separation[J].IEEE Signal Processing, Letters,1997,4(4):112-114.
[18] Amari S I, Cardoso J F. Blind Source Separation: Semi parametric Statistical Approach[J].IEEE Trans on Signal Processing,1997,45(11):2692-2697.
[19] Douglas S C, Cichocki A, Amari S. Multichannel blind separation and deconvolution of sources with arbitrary distributions[J], Neural Networks for Signal Processing, pp:436-445, Sep,1997.
[20] Chen T P, Cichocki A. Stability analysis of adaptive blind source separation[J], Neural Networks, Vol. 10, pp:1345-1351, 1997
[21] Amari S, Kawanabe M. Information geometry of estimating functions in semi-parametric statistical models[J], Bernoulli, pp:29-54, 1997
[22] Amari S, Kawanabe M. Estimating functions in semi-parametric statistical models[J], Institue of Mathematical Statistics, Vol 32, pp:65-81, 1997
[23] Moreau E,Moreau N T. Non-symmetric Contrasts for Source Separation[J].IEEE Trans on Signal Processing,1999,47(8):2241-2252.
[24] Amari S. A theory of adaptive pattern classifiers[J].IEEE Trane, Electronic Computers,1967,16:299-307.
[25] Amari S. Natural gradient works efficiently in learning [J]. Neural computation, 1998,10:251-276.
[26] Tong L, Inouye Y,R Liu. Waveform-preserving blind estimation of multiple sources[J]. IEEE Trans. Signal processing,1993,41(7):2461-2470
[27] Cardeso J F. High-Order Contrasts for Independent Component Analysis[J]. Neural computation, 1999, 11, 157-192.
[28] Hyvarinen A, Oja E. Independent Component Analysis: Algorithms and Applications[J], Neural computation,2000, 13(4-5): 411-430.
[29] Cruces S, Castedo L, Cichocki A. Robust blind source separation algorithms using cumulants[J], Neural computation,2002, 49:87-118
[30] Dam H H, Nordholm S, Low S Y, Cantoni A. Blind Signal Separation Using Steepest Descent Method[J]. IEEE Trans on Signal Processing, 2007,55(8):4198-4207
[31] Jafari M G, Alty S R, Chambers J A. New natural gradient algorithm for cyclostationary sources[J]. IEE Proc-Vis. Image Signal Process, 2004,151(1):62-68
[32] Cruces A S, Cichocki A,Castedo R I . An iterative inversion approach to blindsource separation[J], IEEE Trans. Neural Networks, 2000,11(6):1423-1437
[33] Cardoso J F. Higher order contrasts for independent component analysis[J]. Neural Computation, Vol 11, No. 1, pp:157-192, Jun 1,1999
[34] Amari S, Chen T P, Cichocki A. Non-holonomic Orthogonal Learning Algorithms for Blind Source Separation[J], Neural Computation, Vol. 12, No. 6, pp: 1463-1484, Jun 2000
[35] Shereshevski Y, Yeredor A, Messer H. Super-efficiency in blind source separation of Symmetricheavy-tailed sources[J]. 11th IEEE signal Procssing, pp: 78-81, 2001
[36] Konstantions I D, Theophilos P. Blind Separation of reflections using the image mixtures ratio[J].IEEE image Processing, 2005,2:1034-1037
[37]张贤达,保铮.盲信号分离[J].电子学报,2001,29(12A):1766-1771.
[38]张贤达.盲信号处理几个关键问题的研究[J].深圳大学学报理工版, 2004, 21(3), 196-200
[39]张贤达.矩阵分析与应用[M].北京:清华大学出版社, 2004. pp: 500-589
[40]朱孝龙,张贤达.基于选优估计函数的盲信号分离[J].西安电子科技大学学报,2003,30(3):335-339.
[41]李小军,朱孝龙,张贤达.盲信号分离研究分类与展望[J].西安电子科技大学学报(自然科学版),2004,31(3),399-404.
[42]刘琚,何振亚.盲源分离和盲反卷积[J].电子学报,2002,30(4):570-576.
[43]丁志中,叶中付.基于负熵准则分离方法的剖析与研究[J].系统仿真学报, 2007, 19(13), 2999-3004
[44]牛奕龙,王英民,王毅.一种改进的自适应不完整自然梯度盲源分离算法[J].模式识别与人工智能,2006, 19(5), 667-673.
[45]孙守宇,郑君里,吴德伟.基于自然梯度算法的盲信源分离研究[J].空军工程大学学报(自然科学版), 2003, 4(3):50-54
[46]朱孝龙,张贤达.基于奇异值分解的超定盲信号分离[J].电子与信息学报, 2004,26(3):337-343
[47]冶继民,张贤达,金海红.超定盲信号分离的半参数统计方法[J].电波科学学报, 2006, 21(3):331-336
[48]李小军,楼顺天,张贤达.基于估计概率密度函数的独立分量分析方法[J].西安电子科技大学学报(自然科学版), 2005,32(4):574-578
[49]冯大政,保铮,张贤达.信号盲分离问题多阶段分解算法[J].自然科学进展, 2002,12(3):324-328
[50] The fast-ICA MATAB packer[E]. Available athttp://www.cis.hut.fi/projects/ica/fastica/1998
[51]李雪松,张天骐,杨柳飞等.基于图像单调特性的盲源分离算法[J].计算机应用研究,Dec.2008,Vol.25, No.12,pp: 3665-3668
[52]李雪松,张天骐,杨柳飞等.DS-UWB信号伪码周期的谱估计[J].系统仿真学报(已录用)
[53]史习智.盲信号处理――理论与实践[M].上海交通大学出版社,2008年
[54]朱秀昌,刘峰,胡栋.数字图像处理与图像通信[M].北京:北京邮电大学出版社,2002
[55] Bell A. J., Sejnowski T. J. The "Independent Components" of Natural Scenes are Edge Filters[J]. Vision Res., 1997,37(23), pp:3327-3338
[56]李雪松,张天骐,杨柳飞,代少升,“ICA的一种新应用:图像的滤波”,计算机工程,已录用
[57]李雪松,张天骐等,杨柳飞,代少升,“The period estimation of PN sequence using reprocessing spectrum of DS-UWB signal”,2009 ICCSN,已录用
[58]张天骐,李雪松,代少升,杨柳飞,“一种能同时估计直扩信号伪码周期及码片速率的方法”,舰船电子对抗,Jun. 2008, Vol.31, No.3, pp.77-83
[59]张天骐,代少升,李雪松,杨柳飞,“A Neural Network Method to Adaptive Feature Extraction of Weak DS-CDMA Signals,”IEEE 2008 International Congress on Image and Signal Processing(CISP2008), 27-30 May, Sanya, Hainan, China, Vol.5, pp.385-389.
[60]杨柳飞,张天骐,李雪松,代少升,“低信噪比DS-UWB信号伪码的盲估计方法”,通信技术,2008年4月,Vol.41,No.05, pp:32-34
[61]杨柳飞,张天骐,李雪松,代少升,“微弱DS-UWB信号的盲估计方法研究”,现代雷达,2009年2月,Vol.31,No.2, pp:54-58
[62]杨柳飞,张天骐,李雪松,代少升,“DS-UWB信号参数估计的自相关算法”,电视技术,已录用
[63]张天骐,代少升,杨柳飞,李雪松,“Estimate and Track the PN Sequence of Weak DS-SS Signals,”IEEE the 2007 International Conference on Computational Intelligence and Security (CIS2007), 15-19 Dec. 2007, Harbin China, pp.52-56.
[64]张天骐,杨柳飞,代少升,李雪松,“直扩信号伪码周期及码片速率估计的自相关方法,”航天电子对抗,2007年8月, Vol.23, No.4, pp.53-57.
[65]张天骐,代少升,杨柳飞,李雪松,“在残余频偏下微弱直扩信号伪码周期的谱检测,”系统工程与电子技术,录用已投修改稿.
[2] Cichocki A, Amari S. Adaptive Blind Signal and Image Processing[M]. New York:Wiley, 2005.
[3] Hyvarinen A, Karhunen J, Oja E. Independent Component Analysis[M].New York:Wiley,2001.
[4] Comon P. Independent Component Analysis, a New Concept?[J].Signal Processing,1994,36(3):287-314.
[5] Moreau E, Maechi O. High-order Contrasts for Self-adaptive Source Separation[J].Int J of Adaptive Control and Signal Processing,1996,10(1):19-46.
[6] Bell A J, Sejnowski T J. An Information-Maximization Approach to Blind Separation and Blind Deconvolution[J], Neural Computation, Vol. 7, No. 6, pp:1129-1159, Nov, 1995
[7] Cardeso J F. Blind Signal Separation : Statistical Principles[J]. Proc IEEE,1998,86(10):2009-2025.
[8] Cardoso J F. Eigen-structure of the fourth-order cumulant tensor with application to the blind source separation problem[J], Proc. IEEE ICASSP, 1990, 15:2655-2658
[9] Kung S Y, Diamantaras K I, Taur J S. Adaptive Principal Component Extraction(APEX) and Applications[J]. IEEE Trans on Signal Processing, 1994,42(5):1202-1217
[10] Jutten C, Herault J. Blind separation of sources, Part 1: An adaptive algorithm based on neuromimetic architecture[J]. Signal Processing,1991,24:1-10
[11] Amari S, Ciehocki A. Adaptive Blind Signal Processing : Neural Network Approaches[J]. Proc IEEE,1998,86(10):2026-2048.
[12] Hyvarinen A. Survey on Independent Component Analysis[J].Neural Computing,1999,2(1):94-128.
[13] Delfosse N,Loubaton P. Adaptive blind separation of independent sources :A deflation approach[J].Signal Processing,1995, 45:59-83.
[14] Cardoso J F, Laheld B H. Equivariant adaptive source separation[J]. IEEE Transactions Signal Processing, 1996,44(12):3017-30
[15] Xi Ren Cao,Ruey Wen Liu.General Approach to Blind Source Separation[J].IEEE Transactions on Signal Processing,Vol 44,No 3,1996
[16] Amari S,Cichocki A, Yang HH. A new learning algorithm for blind signal separation[J], Advances in neural information processing systems, pp: 757-763,1996
[17] Cardoso J F. Infomax and Maximum Likelihood for Source Separation[J].IEEE Signal Processing, Letters,1997,4(4):112-114.
[18] Amari S I, Cardoso J F. Blind Source Separation: Semi parametric Statistical Approach[J].IEEE Trans on Signal Processing,1997,45(11):2692-2697.
[19] Douglas S C, Cichocki A, Amari S. Multichannel blind separation and deconvolution of sources with arbitrary distributions[J], Neural Networks for Signal Processing, pp:436-445, Sep,1997.
[20] Chen T P, Cichocki A. Stability analysis of adaptive blind source separation[J], Neural Networks, Vol. 10, pp:1345-1351, 1997
[21] Amari S, Kawanabe M. Information geometry of estimating functions in semi-parametric statistical models[J], Bernoulli, pp:29-54, 1997
[22] Amari S, Kawanabe M. Estimating functions in semi-parametric statistical models[J], Institue of Mathematical Statistics, Vol 32, pp:65-81, 1997
[23] Moreau E,Moreau N T. Non-symmetric Contrasts for Source Separation[J].IEEE Trans on Signal Processing,1999,47(8):2241-2252.
[24] Amari S. A theory of adaptive pattern classifiers[J].IEEE Trane, Electronic Computers,1967,16:299-307.
[25] Amari S. Natural gradient works efficiently in learning [J]. Neural computation, 1998,10:251-276.
[26] Tong L, Inouye Y,R Liu. Waveform-preserving blind estimation of multiple sources[J]. IEEE Trans. Signal processing,1993,41(7):2461-2470
[27] Cardeso J F. High-Order Contrasts for Independent Component Analysis[J]. Neural computation, 1999, 11, 157-192.
[28] Hyvarinen A, Oja E. Independent Component Analysis: Algorithms and Applications[J], Neural computation,2000, 13(4-5): 411-430.
[29] Cruces S, Castedo L, Cichocki A. Robust blind source separation algorithms using cumulants[J], Neural computation,2002, 49:87-118
[30] Dam H H, Nordholm S, Low S Y, Cantoni A. Blind Signal Separation Using Steepest Descent Method[J]. IEEE Trans on Signal Processing, 2007,55(8):4198-4207
[31] Jafari M G, Alty S R, Chambers J A. New natural gradient algorithm for cyclostationary sources[J]. IEE Proc-Vis. Image Signal Process, 2004,151(1):62-68
[32] Cruces A S, Cichocki A,Castedo R I . An iterative inversion approach to blindsource separation[J], IEEE Trans. Neural Networks, 2000,11(6):1423-1437
[33] Cardoso J F. Higher order contrasts for independent component analysis[J]. Neural Computation, Vol 11, No. 1, pp:157-192, Jun 1,1999
[34] Amari S, Chen T P, Cichocki A. Non-holonomic Orthogonal Learning Algorithms for Blind Source Separation[J], Neural Computation, Vol. 12, No. 6, pp: 1463-1484, Jun 2000
[35] Shereshevski Y, Yeredor A, Messer H. Super-efficiency in blind source separation of Symmetricheavy-tailed sources[J]. 11th IEEE signal Procssing, pp: 78-81, 2001
[36] Konstantions I D, Theophilos P. Blind Separation of reflections using the image mixtures ratio[J].IEEE image Processing, 2005,2:1034-1037
[37]张贤达,保铮.盲信号分离[J].电子学报,2001,29(12A):1766-1771.
[38]张贤达.盲信号处理几个关键问题的研究[J].深圳大学学报理工版, 2004, 21(3), 196-200
[39]张贤达.矩阵分析与应用[M].北京:清华大学出版社, 2004. pp: 500-589
[40]朱孝龙,张贤达.基于选优估计函数的盲信号分离[J].西安电子科技大学学报,2003,30(3):335-339.
[41]李小军,朱孝龙,张贤达.盲信号分离研究分类与展望[J].西安电子科技大学学报(自然科学版),2004,31(3),399-404.
[42]刘琚,何振亚.盲源分离和盲反卷积[J].电子学报,2002,30(4):570-576.
[43]丁志中,叶中付.基于负熵准则分离方法的剖析与研究[J].系统仿真学报, 2007, 19(13), 2999-3004
[44]牛奕龙,王英民,王毅.一种改进的自适应不完整自然梯度盲源分离算法[J].模式识别与人工智能,2006, 19(5), 667-673.
[45]孙守宇,郑君里,吴德伟.基于自然梯度算法的盲信源分离研究[J].空军工程大学学报(自然科学版), 2003, 4(3):50-54
[46]朱孝龙,张贤达.基于奇异值分解的超定盲信号分离[J].电子与信息学报, 2004,26(3):337-343
[47]冶继民,张贤达,金海红.超定盲信号分离的半参数统计方法[J].电波科学学报, 2006, 21(3):331-336
[48]李小军,楼顺天,张贤达.基于估计概率密度函数的独立分量分析方法[J].西安电子科技大学学报(自然科学版), 2005,32(4):574-578
[49]冯大政,保铮,张贤达.信号盲分离问题多阶段分解算法[J].自然科学进展, 2002,12(3):324-328
[50] The fast-ICA MATAB packer[E]. Available athttp://www.cis.hut.fi/projects/ica/fastica/1998
[51]李雪松,张天骐,杨柳飞等.基于图像单调特性的盲源分离算法[J].计算机应用研究,Dec.2008,Vol.25, No.12,pp: 3665-3668
[52]李雪松,张天骐,杨柳飞等.DS-UWB信号伪码周期的谱估计[J].系统仿真学报(已录用)
[53]史习智.盲信号处理――理论与实践[M].上海交通大学出版社,2008年
[54]朱秀昌,刘峰,胡栋.数字图像处理与图像通信[M].北京:北京邮电大学出版社,2002
[55] Bell A. J., Sejnowski T. J. The "Independent Components" of Natural Scenes are Edge Filters[J]. Vision Res., 1997,37(23), pp:3327-3338
[56]李雪松,张天骐,杨柳飞,代少升,“ICA的一种新应用:图像的滤波”,计算机工程,已录用
[57]李雪松,张天骐等,杨柳飞,代少升,“The period estimation of PN sequence using reprocessing spectrum of DS-UWB signal”,2009 ICCSN,已录用
[58]张天骐,李雪松,代少升,杨柳飞,“一种能同时估计直扩信号伪码周期及码片速率的方法”,舰船电子对抗,Jun. 2008, Vol.31, No.3, pp.77-83
[59]张天骐,代少升,李雪松,杨柳飞,“A Neural Network Method to Adaptive Feature Extraction of Weak DS-CDMA Signals,”IEEE 2008 International Congress on Image and Signal Processing(CISP2008), 27-30 May, Sanya, Hainan, China, Vol.5, pp.385-389.
[60]杨柳飞,张天骐,李雪松,代少升,“低信噪比DS-UWB信号伪码的盲估计方法”,通信技术,2008年4月,Vol.41,No.05, pp:32-34
[61]杨柳飞,张天骐,李雪松,代少升,“微弱DS-UWB信号的盲估计方法研究”,现代雷达,2009年2月,Vol.31,No.2, pp:54-58
[62]杨柳飞,张天骐,李雪松,代少升,“DS-UWB信号参数估计的自相关算法”,电视技术,已录用
[63]张天骐,代少升,杨柳飞,李雪松,“Estimate and Track the PN Sequence of Weak DS-SS Signals,”IEEE the 2007 International Conference on Computational Intelligence and Security (CIS2007), 15-19 Dec. 2007, Harbin China, pp.52-56.
[64]张天骐,杨柳飞,代少升,李雪松,“直扩信号伪码周期及码片速率估计的自相关方法,”航天电子对抗,2007年8月, Vol.23, No.4, pp.53-57.
[65]张天骐,代少升,杨柳飞,李雪松,“在残余频偏下微弱直扩信号伪码周期的谱检测,”系统工程与电子技术,录用已投修改稿.