基于脑电信号的警觉度估计研究
|
摘要
警觉度(Vigilance)通常定义为，对外界刺激长时间的保持注意力和警惕性的能力。在日常的工作和生活中,许多人机交互系统需要操作人员保持一定的警觉度。尤其是汽车驾驶员，他们一旦出现警觉度下降的情况，就可能会造成非常严重的后果。因此，研究警觉度监测算法，对驾驶员的疲劳状态进行实时监测，是车辆辅助安全驾驶领域的一个重要课题，对解决交通安全问题具有重要的现实意义。由于脑电信号是疲劳检测的“金标准”，以及便携式脑电采集设备商用化的出现，使得脑电采集越来越便捷。因此，本论文将重点进行基于脑电的警觉度监测方法研究，具体研究问题包括：脑电信号在线降噪去伪迹算法，警觉度关键脑区定位，脑电疲劳特征快速提取算法、在线脑电特征过滤平滑算法，高鲁棒性的脑电特征降维算法，高效的在线警觉度监测算法以及警觉度自动标注算法等。目的是结合便携式脑电采集系统，为将来开发基于脑电等生理信号的警觉度监测系统提供理论和基本算法。
本论文的主要贡献和创新点体现在以下几个方面：
1.脑电信号在线降噪去伪迹。本文提出了自适应在线降噪去伪迹算法，能自适应跟踪随时间变化的信号源，并具备伪迹源自动识别功能，且能在线自动重构出不含伪迹成分的脑电信号，以解决脑电采集过程中的噪音伪迹干扰问题。该算法的伪迹识别精度在90%以上，已达到人工标注的水平，相比于其他自动去伪迹算法，该算法能很好的解决伪迹信号源随时间变化的问题，并且支持在线工作模式。
2.警觉度关键脑区定位。本文提出了基于脑电信号的警觉度相关同步去同步概念，基于该原理，开发了三种警觉度关键脑区定位算法，分别是基于线性相关系数的定位方法，基于ICA算法的定位方法和基于CSP算法的定位方法，并最终定位出枕叶附近为警觉度关键脑区。从警觉度关键脑区采集脑电信号可有效降低警觉度无关脑电信号的影响，并可以简化脑电采集过程，增强脑电警觉度分析系统的实用性。
3.脑电疲劳特征提取。对对数形式的脑电频谱特征，本文给出了物理层面的解释，进而定义了脑电的微分熵特征。提出了脑电特征中的乘性噪音问题，并给出了不同脑电特征的乘性噪音消除方案。同时，结合警觉度实验数据，对自回归系数，能量谱，分形维数，样本熵以及微分熵这五种脑电特征，从原理和性能两个层面进行了详细的对比分析。最终结果显示，微分熵特征的警觉度表征能力最强，且性能最稳定。
4.脑电特征在线过滤。本文提出了基于线性动力系统模型的脑电特征过滤算法，用于消除脑电数据中的警觉度无关脑电特征。同时，结合警觉度实验数据，从原理和性能两个方面与传统的滑动平均滤波器算法进行了对比分析。结果显示，该算法能更加充分合理的利用已观测到的全部脑电数据，更好的消除警觉度无关脑电特征的影响，提高警觉度的估计精度。
5.鲁棒性的脑电特征降维。由于脑电特征中含有较强的噪音，传统的特征降维算法如PCA，在受到噪音干扰后会导致计算出的主成分不准确，从而影响后续的警觉度估计性能。为了解决特征降维过程中的噪音问题，本文引入了具有抗噪能力的特征降维算法：Robust PCA，并基于具体的实验数据，进行了对比分析。结果显示，Robust PCA算法相比于标准PCA算法，在处理含有较多噪音的脑电特征时，确实能有效提高警觉度估计系统的性能。
6.高效的警觉度估计模型。本文提出了用于在线警觉度估计的Larsen-ELM算法，相比于SVM回归模型，该算法在保证警觉度估计准确率的同时，具有高效的模型参数训练效率，对处理大规模的脑电警觉度数据，具有较高的实用价值。
7.基于脑电信号的警觉度自动标注算法。传统警觉度标注算法通常代价过高，自动化性能欠佳，且无法有效处理大规模的数据集。因此，本文提出了直接基于脑电数据的离散警觉度标注算法和连续警觉度标注算法。由于脑电特征随警觉度变化而变化，该类型的算法能直接利用脑电特征的分布规律通过聚类算法和流形学习算法进行警觉度的自动标注。直接基于脑电数据的标注结果，与传统的基于警觉度相关任务行为的标注结果相比，同样能准确的标注出警觉度变化的整体走势，只是在细节方面存在较小的误差。由于该类型的算法是完全自动化实现，且无需训练数据集，因此，该类型的算法能有效的提高脑电警觉度数据集的标注效率。
最后，我们设计了一套完整的脑电警觉度实验方案，采集了大量的脑电警觉度实验数据，在充分且具体的数据支撑下，从实验结果和算法机理两个方面，对警觉度分析过程中各个环节的不同算法，进行了系统性的对比分析，给出了各个分析环节的推荐算法使用方案；实验结果显示，我们提出的警觉度监测方法，确实能准确可靠的实现在线警觉度估计功能和离线警觉度标注功能。
Vigilance usually is defined as the ability to maintain focus of attention and to remain alertto stimuli for prolonged periods of time. In our daily lives, for many human-machine interac-tion systems, the operators should retain their vigilance above a constant level. Especially forcar drivers, losing vigilance may cause some serious traffic accidents. Therefore, research onvigilance monitoring algorithms and monitoring the driver’s vigilance level in real-time is an im-portant topic for safe driving in vehicle auxiliary, and has important practical significance to solvethe problem of traffic safety. Because EEG signal is the “gold standard” for vigilance detection,and the emergence of commercial portable EEG recording equipment makes EEG recording moreconvenient. In this thesis, we will focus on the key methods for EEG-based vigilance monitor-ing. The research topic include: on-line noise reduction and artifact removal algorithms for EEGpreprocessing, vigilance-related critical brain region location, efficient vigilance-related featureextraction algorithm, on-line EEG feature filtering algorithm, robust vigilance feature dimensionreduction algorithm, efficient on-line vigilance monitor algorithm, and vigilance automatic annota-tion algorithm. The main goal is, together with the portable EEG recording system, to provide keytechnologies and theories for future EEG and other physiology signals-based vigilance monitoringsystem.
The main contributions and innovations of this thesis are listed as below:
1. EEG on-line noise reduction and artifact removal. An adaptive on-line noise reduction andartifact removal algorithm is proposed, which can adaptively track the time-varying signalsource, automatically recognize the artifact source, and on-line reconstruct the EEG signalswithout artifact. This algorithm can solve the problem of noise or artifact interference inEEG recording. The artifact recognition precision is more than90%, which has reached thelevel of manual artifact annotation. Compared with other artifact removal algorithms, thisalgorithm can solve the problem of artifact source changing with time, and can support theon-line analysis mode.
2. vigilance-related critical brain region location. The concept of EEG-based vigilance-relatedsynchronization and desynchronization is proposed. Based on this concept, three algo-rithms for vigilance-related critical brain region location is designed, which are correlation coefficient-based location algorithm, ICA-based location algorithm, and CSP-based locationalgorithm. And critical brain region is found, which is near the occipital lobe. RecordingEEG signals directly from the critical brain region can dramatically reduce the influenceof vigilance-unrelated EEG signals, simplify the EEG recording process, and enhance thepracticability of EEG-based vigilance analysis system.
3. Vigilance-related feature extraction. For the logarithmic form of the EEG spectral feature,we have presented an explanation from the physical aspect, and defined it as the differentialentropy feature of EEG. The multiplicative noise problem of EEG feature is drawn out, andthe corresponding noise reduction method also is proposed. Based on the EEG-vigilanceexperiment data set, a systematic comparative study on performance and algorithm mech-anism is conducted for five kinds of EEG features, autoregressive coefficient, energy spec-trum, fractal dimension, sample entropy, and differential entropy. Experiment results showthat, differential entropy is the most accurate and stable EEG feature to reflect the vigilancechanges.
4. On-line EEG feature filtering. A linear dynamical system (LDS)-based EEG feature filteringalgorithm is proposed, which is used to on-line remove the vigilance-unrelated EEG featurefrom the original EEG feature. Based on the EEG-vigilance experiment data set, a compar-ative study on performance and algorithm mechanism is conducted between the LDS-basedfiltering algorithm and the traditional moving average filter. Experiment results show that,LDS-based filtering algorithm can more fully use the observed EEG data, is better to elimi-nate the influence of the vigilance-unrelated EEG feature, and finally improve the vigilanceestimation accuracy.
5. Robust vigilance feature dimension reduction. Usually the noise in EEG feature is muchstronger, if we directly use the ordinary PCA algorithm for feature dimension reduction,with the influence of noise, the extracted principal components will be skewed, then the vig-ilance estimation accuracy will be reduced. To solve the noise problem of feature dimensionreduction, noised unsensitive dimension reduction algorithm is introduced, which is RobustPCA. A comparative study is conducted based on experiment data. Experiment result showthat, compared with the ordinary PCA, Robust PCA can effectively improve the performanceof the vigilance estimation system when the EEG features containing lots of noise.
6. Efficient on-line vigilance estimation algorithm. An on-line vigilance estimation algorithmnamed Larsen-ELM is proposed. Compared with SVM regression model, Larsen-ELM candramatically speed up the model training process while still achieve a similar vigilance es-timation accuracy. For large scale EEG-vigilance data analysis, Larsen-ELM has a highpractical value.
7. EEG-based automatic vigilance annotation algorithm. The traditional vigilance annotationmethods usually have a very high cost, very poor automation capability, and can not handlethe large scale data set. Therefore, two kinds of EEG-based vigilance annotation algorithms,discrete annotation algorithm and continuous annotation algorithm are proposed. Since theEEG feature is affected by the vigilance changes, the proposed algorithms can directly usedthe distribution information of the EEG feature combine with the clustering algorithm ormanifold learning algorithm for automatic vigilance annotation. Compared with the tradi-tional annotation algorithm, the proposed algorithms also can accurately annotate the trendof vigilance changes, but only have small errors in the detail vigilance annotation. As theproposed algorithms can automatically annotate the EEG data, and do not need training data,by using these algorithms, the efficiency of vigilance annotation on EEG-vigilance data setwill be dramatically improved.
Finally, we have designed an EEG-vigilance experiment, and have collected lots of EEG-vigilance experiment data. With the support of adequate experiment data and from the aspectsof experiment results and algorithm mechanism, a systematic comparative study on system per-formance is conducted for different algorithms on every vigilance analysis sub process. Then therecommendation algorithms for every vigilance analysis sub process have been raised. Experimentresults show that, the proposed method for EEG-based vigilance estimation, can indeed achieveaccurate and reliable on-line vigilance estimation and off-line vigilance annotation.
本论文的主要贡献和创新点体现在以下几个方面：
1.脑电信号在线降噪去伪迹。本文提出了自适应在线降噪去伪迹算法，能自适应跟踪随时间变化的信号源，并具备伪迹源自动识别功能，且能在线自动重构出不含伪迹成分的脑电信号，以解决脑电采集过程中的噪音伪迹干扰问题。该算法的伪迹识别精度在90%以上，已达到人工标注的水平，相比于其他自动去伪迹算法，该算法能很好的解决伪迹信号源随时间变化的问题，并且支持在线工作模式。
2.警觉度关键脑区定位。本文提出了基于脑电信号的警觉度相关同步去同步概念，基于该原理，开发了三种警觉度关键脑区定位算法，分别是基于线性相关系数的定位方法，基于ICA算法的定位方法和基于CSP算法的定位方法，并最终定位出枕叶附近为警觉度关键脑区。从警觉度关键脑区采集脑电信号可有效降低警觉度无关脑电信号的影响，并可以简化脑电采集过程，增强脑电警觉度分析系统的实用性。
3.脑电疲劳特征提取。对对数形式的脑电频谱特征，本文给出了物理层面的解释，进而定义了脑电的微分熵特征。提出了脑电特征中的乘性噪音问题，并给出了不同脑电特征的乘性噪音消除方案。同时，结合警觉度实验数据，对自回归系数，能量谱，分形维数，样本熵以及微分熵这五种脑电特征，从原理和性能两个层面进行了详细的对比分析。最终结果显示，微分熵特征的警觉度表征能力最强，且性能最稳定。
4.脑电特征在线过滤。本文提出了基于线性动力系统模型的脑电特征过滤算法，用于消除脑电数据中的警觉度无关脑电特征。同时，结合警觉度实验数据，从原理和性能两个方面与传统的滑动平均滤波器算法进行了对比分析。结果显示，该算法能更加充分合理的利用已观测到的全部脑电数据，更好的消除警觉度无关脑电特征的影响，提高警觉度的估计精度。
5.鲁棒性的脑电特征降维。由于脑电特征中含有较强的噪音，传统的特征降维算法如PCA，在受到噪音干扰后会导致计算出的主成分不准确，从而影响后续的警觉度估计性能。为了解决特征降维过程中的噪音问题，本文引入了具有抗噪能力的特征降维算法：Robust PCA，并基于具体的实验数据，进行了对比分析。结果显示，Robust PCA算法相比于标准PCA算法，在处理含有较多噪音的脑电特征时，确实能有效提高警觉度估计系统的性能。
6.高效的警觉度估计模型。本文提出了用于在线警觉度估计的Larsen-ELM算法，相比于SVM回归模型，该算法在保证警觉度估计准确率的同时，具有高效的模型参数训练效率，对处理大规模的脑电警觉度数据，具有较高的实用价值。
7.基于脑电信号的警觉度自动标注算法。传统警觉度标注算法通常代价过高，自动化性能欠佳，且无法有效处理大规模的数据集。因此，本文提出了直接基于脑电数据的离散警觉度标注算法和连续警觉度标注算法。由于脑电特征随警觉度变化而变化，该类型的算法能直接利用脑电特征的分布规律通过聚类算法和流形学习算法进行警觉度的自动标注。直接基于脑电数据的标注结果，与传统的基于警觉度相关任务行为的标注结果相比，同样能准确的标注出警觉度变化的整体走势，只是在细节方面存在较小的误差。由于该类型的算法是完全自动化实现，且无需训练数据集，因此，该类型的算法能有效的提高脑电警觉度数据集的标注效率。
最后，我们设计了一套完整的脑电警觉度实验方案，采集了大量的脑电警觉度实验数据，在充分且具体的数据支撑下，从实验结果和算法机理两个方面，对警觉度分析过程中各个环节的不同算法，进行了系统性的对比分析，给出了各个分析环节的推荐算法使用方案；实验结果显示，我们提出的警觉度监测方法，确实能准确可靠的实现在线警觉度估计功能和离线警觉度标注功能。
Vigilance usually is defined as the ability to maintain focus of attention and to remain alertto stimuli for prolonged periods of time. In our daily lives, for many human-machine interac-tion systems, the operators should retain their vigilance above a constant level. Especially forcar drivers, losing vigilance may cause some serious traffic accidents. Therefore, research onvigilance monitoring algorithms and monitoring the driver’s vigilance level in real-time is an im-portant topic for safe driving in vehicle auxiliary, and has important practical significance to solvethe problem of traffic safety. Because EEG signal is the “gold standard” for vigilance detection,and the emergence of commercial portable EEG recording equipment makes EEG recording moreconvenient. In this thesis, we will focus on the key methods for EEG-based vigilance monitor-ing. The research topic include: on-line noise reduction and artifact removal algorithms for EEGpreprocessing, vigilance-related critical brain region location, efficient vigilance-related featureextraction algorithm, on-line EEG feature filtering algorithm, robust vigilance feature dimensionreduction algorithm, efficient on-line vigilance monitor algorithm, and vigilance automatic annota-tion algorithm. The main goal is, together with the portable EEG recording system, to provide keytechnologies and theories for future EEG and other physiology signals-based vigilance monitoringsystem.
The main contributions and innovations of this thesis are listed as below:
1. EEG on-line noise reduction and artifact removal. An adaptive on-line noise reduction andartifact removal algorithm is proposed, which can adaptively track the time-varying signalsource, automatically recognize the artifact source, and on-line reconstruct the EEG signalswithout artifact. This algorithm can solve the problem of noise or artifact interference inEEG recording. The artifact recognition precision is more than90%, which has reached thelevel of manual artifact annotation. Compared with other artifact removal algorithms, thisalgorithm can solve the problem of artifact source changing with time, and can support theon-line analysis mode.
2. vigilance-related critical brain region location. The concept of EEG-based vigilance-relatedsynchronization and desynchronization is proposed. Based on this concept, three algo-rithms for vigilance-related critical brain region location is designed, which are correlation coefficient-based location algorithm, ICA-based location algorithm, and CSP-based locationalgorithm. And critical brain region is found, which is near the occipital lobe. RecordingEEG signals directly from the critical brain region can dramatically reduce the influenceof vigilance-unrelated EEG signals, simplify the EEG recording process, and enhance thepracticability of EEG-based vigilance analysis system.
3. Vigilance-related feature extraction. For the logarithmic form of the EEG spectral feature,we have presented an explanation from the physical aspect, and defined it as the differentialentropy feature of EEG. The multiplicative noise problem of EEG feature is drawn out, andthe corresponding noise reduction method also is proposed. Based on the EEG-vigilanceexperiment data set, a systematic comparative study on performance and algorithm mech-anism is conducted for five kinds of EEG features, autoregressive coefficient, energy spec-trum, fractal dimension, sample entropy, and differential entropy. Experiment results showthat, differential entropy is the most accurate and stable EEG feature to reflect the vigilancechanges.
4. On-line EEG feature filtering. A linear dynamical system (LDS)-based EEG feature filteringalgorithm is proposed, which is used to on-line remove the vigilance-unrelated EEG featurefrom the original EEG feature. Based on the EEG-vigilance experiment data set, a compar-ative study on performance and algorithm mechanism is conducted between the LDS-basedfiltering algorithm and the traditional moving average filter. Experiment results show that,LDS-based filtering algorithm can more fully use the observed EEG data, is better to elimi-nate the influence of the vigilance-unrelated EEG feature, and finally improve the vigilanceestimation accuracy.
5. Robust vigilance feature dimension reduction. Usually the noise in EEG feature is muchstronger, if we directly use the ordinary PCA algorithm for feature dimension reduction,with the influence of noise, the extracted principal components will be skewed, then the vig-ilance estimation accuracy will be reduced. To solve the noise problem of feature dimensionreduction, noised unsensitive dimension reduction algorithm is introduced, which is RobustPCA. A comparative study is conducted based on experiment data. Experiment result showthat, compared with the ordinary PCA, Robust PCA can effectively improve the performanceof the vigilance estimation system when the EEG features containing lots of noise.
6. Efficient on-line vigilance estimation algorithm. An on-line vigilance estimation algorithmnamed Larsen-ELM is proposed. Compared with SVM regression model, Larsen-ELM candramatically speed up the model training process while still achieve a similar vigilance es-timation accuracy. For large scale EEG-vigilance data analysis, Larsen-ELM has a highpractical value.
7. EEG-based automatic vigilance annotation algorithm. The traditional vigilance annotationmethods usually have a very high cost, very poor automation capability, and can not handlethe large scale data set. Therefore, two kinds of EEG-based vigilance annotation algorithms,discrete annotation algorithm and continuous annotation algorithm are proposed. Since theEEG feature is affected by the vigilance changes, the proposed algorithms can directly usedthe distribution information of the EEG feature combine with the clustering algorithm ormanifold learning algorithm for automatic vigilance annotation. Compared with the tradi-tional annotation algorithm, the proposed algorithms also can accurately annotate the trendof vigilance changes, but only have small errors in the detail vigilance annotation. As theproposed algorithms can automatically annotate the EEG data, and do not need training data,by using these algorithms, the efficiency of vigilance annotation on EEG-vigilance data setwill be dramatically improved.
Finally, we have designed an EEG-vigilance experiment, and have collected lots of EEG-vigilance experiment data. With the support of adequate experiment data and from the aspectsof experiment results and algorithm mechanism, a systematic comparative study on system per-formance is conducted for different algorithms on every vigilance analysis sub process. Then therecommendation algorithms for every vigilance analysis sub process have been raised. Experimentresults show that, the proposed method for EEG-based vigilance estimation, can indeed achieveaccurate and reliable on-line vigilance estimation and off-line vigilance annotation.
引文
[1]过秀成.道路安全学[M].南京:东南大学出版社,2001:31–32.
[2] DAWSON D, LAMOND N, DONKIN K. Quantitative similarity between the cognitive psy-chomotor performance decrement associated with sustained wakefulness and alcohol intoxi-cation[C]. Proceedings of the Third International Conference on Fatigue and Transportation.1998:44–45.
[3]陈伟.疲劳驾驶:法国人的防范之策[J].社会发展,2000,10:43–44.
[4] KENNETH P, WRIGHT J, JOSEPH H. Relationship between alertness,performance,and bodytemperature in humans[J]. American Journal of Physiology,2002,283(6):R1370–R1377.
[5] CAJOCHEN C, KHALSA S, WYATT J. EEG and ocular correlates of circadian melatoninphase and human performance decrements during sleep loss[J]. American Journal of Phys-iology,1999,277(3):R640–R649.
[6] DINGES D F, MALLIS M, MAISLIN G, et al. Evaluation of techniques for ocular measure-ment as an index of fatigue and the basis for alertness management[R]. National HighwayTraffic Safety Administration, USA,1998.
[7] JI Q, YANG X. Real-time eye, gaze, and face pose tracking for monitoring driver vigi-lance[J]. Real-Time Imaging,2002,8(5):357–377.
[8] BOVERIS S. Driver fatigue monitoring technologies and future ideas[C]. Proceedings ofAWAKE Road Safety Workshop. Balocco, ItalyElsevier Science Ltd,2004.
[9] MATTHEWS R, MCDONALD N J, HERVIEUX P, et al. A wearable physiological sensorsuite for unobtrusive monitoring of physiological and cognitive state[C]. Proc. IEEE Int.Conf. of Engineering in Medicine and Biology Society.2007:5276–5281.
[10] WANG X W, SHI L C, LU B L. A Survey on Technology of Dry Electrodes for EEGRecording[J]. Chinese Journal of Biomedical Engineering,2010,29:777–784.
[11] PFURTSCHELLER G, ARANIBAR A. Event-related cortical desynchronization detected bypower measurements of scalp EEG[J]. Electroencephalography Clinical Neurophysiology,1977,42(6):817–826.
[12] MATOUSEK M, PETERSEN I. A method for assessing alertness fluctuations from EEGspectra[J]. Electroencephalography Clinical Neurophysiology,1983,55(1):108–113.
[13]李文磊.基于DSP的疲劳驾驶实时监测系统研究[D].南京:南京理工大学,2007.
[14] HYOKI K, SHIGETA M, TSUNOB N. Quantitative electro-oculography and electro-en-cephalography as indices of alertness[J]. Electroencephalography and clinical Neurophysi-ology,1998,106(3):213–219.
[15] MAGOSSO E, PROVINI F, MONTAGNA P. A wavelet based method for automatic detec-tion of slow eye movements: A pilot study[J]. Medical Engineering and Physics,2006,28(9):860–875.
[16] MAGOSSO E, URSINO M, ZANIBONI A. Visual and computer-based detection of sloweye movements in overnight and24-h EOG recordings[J]. Clinical Neurophysiology,2007,118(5):1122–1133.
[17] ATIENZA M, CANTERO J L, STICKGOLD R. Eyelid movements measured by Nightcappredict slow eye movements during quiet wakefulness in humans[J]. Journal of Sleep Re-search,2004,13(1):25–29.
[18] KAIDA K, AKERSTED T, KECKLUND G. Use of Subjective and Physiological Indicatorsof Sleepiness to Predict Performance during a Vigilance Task[J]. Industrial Health,2007,45(4):520–526.
[19] JAMMES B, SHARABTY H, ESTEVE D. Automatic EOG analysis: A first step toward auto-matic drowsiness scoring during wake-sleep transitions[J]. Somnologie,2008,12(3):227–232.
[20] OKEN B, SALINSKY M, ELSAS S. Vigilance, alertness, or sustained attention: physiologi-cal basis and measurement[J]. Clinical Neurophysiology,2006,117:1885–1901.
[21]毛喆.基于驾驶员生理特征分析的驾驶疲劳状态识别方法研究[D].武汉:武汉理工大学,2006.
[22] KUBICKI S, HERRMANN W, FICHTE K, et al. Reflections on the topics: EEG frequencybands and regulation of vigilance[J]. Pharmakopsychiatr Neuropsychopharmakol,1979,12(2):237–245.
[23] MAKEIG S, INLOW M. Lapses in alertness: coherence of fluctuations in performance andEEG spectrum[J]. Electroencephalography and Clinical Neurophysiology,1993,86(1):23–35.
[24] MAKEIG S. Using Feedforward Neural Networks to Monitor Alertness from Changes inEEG Correlation and Coherence[J]. Advances in Neural Information Processing Systems,1996,8:931–937.
[25] T.P.JUNG, MAKEIG S, STENSMO M, et al. Estimating Alertness from the EEG PowerSpectrum[J]. IEEE Transactions on Biomedical Engineering,1997,44(1):60–69.
[26] HUANG R, JUNG T, DELORME A, et al. Tonic and phasic electroencephalographic dynam-ics during continuous compensatory tracking[J]. NeuroImage,2008,39(4):1896–1909.
[27] LIN C, WU R, LIANG S, et al. EEG-Based Drowsiness Estimation for Safety Driving UsingIndependent Component Analysis[J]. IEEE Transaction on Circuits and Systems-I,2005,52(12):2726–2738.
[28] LIN C, KO L, CHUNG I, et al. Adaptive EEG-Based Alertness Estimation System by UsingICA-Based Fuzzy Neural Networks[J]. IEEE Transactions on Circuits and Systems-I,2006,53(11):2469–2476.
[29] LIN C, HUANG K, CHAO C, et al. Tonic and phasic EEG and behavioral changes inducedby arousing feedback[J]. NeuroImage,2010,52(2):633–642.
[30] LIN C, KO L, CHIOU J, et al. Noninvasive neural prostheses using mobile and wirelessEEG[J]. Proc. IEEE,2008,96(7):1167–1183.
[31] CHI Y M, CAUWENBERGHS G. Micropower Non-contact EEG Electrode with ActiveCommon-Mode Noise Suppression and Input Capacitance Cancellation[J]. Proc. IEEE Int.Conf. of Engineering in Medicine and Biology Society,2009:4218–4221.
[32] DAVIDSON P R, JONES R D, PEIRIS M T R. EEG-Based Lapse Detection With HighTemporal Resolution[J]. IEEE Transactions on Biomedical Engineering,2007,54(5):832–839.
[33]赵仑. ERP实验教程[M].天津:天津社会科学院出版社,2004:11–12.
[34] NORDBY H, HUGDAHL K, STICKGOLD R, et al. Event-related potentials (ERPs) to deviantauditory stimuli during sleep and waking[J]. NeuroReport,1996,7:1082–1086.
[35]傅佳伟,石立臣,吕宝粮.基于EEG的警觉度分析与估计研究综述[J].中国生物医学工程学报,2009,28(4):589–596.
[36] RICHMAN J S, MOORMAN J R. Physiological time-series analysis using approximate en-tropy and sample entropy[J]. American Journal of Physiology,2000,278(6):H2039–H2049.
[37] ESTELLER R, VACHTSEVANOS G, ECHAUZ J, et al. A comparison of waveform fractal di-mension algorithms[J]. IEEE Transactions on Circuits and Systems I: Fundamental Theoryand Applications,2001,48(2):177–183.
[38] PARDEY J, ROBERTS S, TARASSENKO L. A review of parametric modelling techniquesfor EEG analysis[J]. Medical Engineering&Physics,1996,18(1):2–11.
[39] FATOURECHI M, BASHASHATI A, WARD R K, et al. EMG and EOG artifacts in braincomputer interface systems: A survey[J]. Clinical Neurophysiology,2007,118:480–494.
[40] GRATTON G. Dealing with artifacts: the EOG contamination of the event-related brainpotential[J]. Behav Res Methods, Instrum, Comput,1998,30:44–53.
[41] MILLAN J, FRANZE M, MOURINO J, et al. Relevant EEG features for the classi?cation ofspontaneous motor-related tasks[J]. Biol Cybern,2002,86:89–95.
[42] GOTMAN J, SKUCE D, THOMPSON C, et al. Clinical applications of spectral analysis andextraction of features from electroencephalograms with slow waves in adult patients[J]. BiolCybern,1973,35:225–235.
[43] CROFT R, CHANDLER J, BARRY R, et al. EOG correction: a comparison of four meth-ods[J]. Psychophysiology,2005,42:16–24.
[44] LAGERLUND T, SHARBROUGH F, BUSACKER N. Spatial filtering of multichannel elec-troencephalographic recordings through principal component analysis by singular value de-composition[J]. Journal of clinical neurophysiology,1997,14:73–82.
[45] JUNG T, MAKEIG S, WESTERFIELD M, et al. Removal of eye activity artifacts from visualevent-related potentials in normal and clinical subjects[J]. Clinical Neurophysiology,2000,111:1745–1758.
[46] LACHAUX J, RODRIGUEZ E, MARTINERIE J, et al. Measuring Phase Synchrony in BrainSignals[J]. Human Brain Mapping,1999,8:194–208.
[47] PEIRIS M, JONES R, DAVIDSON P, et al. Identification of Vigilance Lapses usingEEG/EOG by Expert Human Raters[C]. Proc. IEEE Int. Conf. of Engineering in Medicineand Biology Society.2005:5735–5737.
[48] JOHNS M, TUCKER A, CHAPMAN R. A New Method for Monitoring the Drowsiness ofDrivers[C]. Proc. Int. Conf. on Fatigue Management in Transportation Operations.2005:11–15.
[49] GONCHAROVA I I, MCFARLAND D J, VAUGHAN T M, et al. EMG contamination of EEG:spectral and topographical characteristics[J]. Clinical Neurophysiology,2003,114:1580–1593.
[50] BLUME W T, KAIBARA M, YOUNG G B. Atlas of Adult Electroencephalography[M].Lippincott Williams&Wilkins,2002.
[51] BARLOW J S. Artifact processing (rejection and minimization) in EEG data processing[J].Clinical Neurophysiology,1986,2:15–62.
[52] MA J X, SHI L C, LIANG LU B. Vigilance Estimation by Using ElectrooculographicFeatures[C].32nd Annual International Conference of the IEEE Engineering in Medicineand Biology Society (EMBC10).2010:6591–6594.
[53] CAI H Y, MA J X, SHI L C, et al. A novel method for EOG features extraction from theforehead[C].33rd Annual International Conference of the IEEE Engineering in Medicineand Biology Society (EMBC11).2011:3075–3078.
[54] ANDERER P, ROBERTS S, SCHLOGL A, et al. Artifact Processing in Computerized Analy-sis of Sleep EEG-A Review[J]. Neuropsychobiology,1999,40:150–157.
[55] OCHOA C J, POLICH J. P300and blink instructions[J]. Clinical Neurophysiology,2000,111:93–98.
[56] VAN DE VELDE M, VAN ERP G, CLUITMANS P J. Detection of muscle artefact in the nor-mal human awake EEG[J]. Electroencephalography and Clinical Neurophysiology,1998,107:149–158.
[57] VAN DE VELDE M, VAN ERP G, CLUITMANS P J. Context related artefact detection inprolonged EEG recordings[J]. Computer Methods and Programs in Biomedicine,1999,60:183–196.
[58] CHOI S, CICHOCKI A, PARK H M, et al. Blind source separation and independent compo-nent analysis: a review[J]. Neural Inf Process-Lett Rev,2005,6:1–57.
[59] HYVARINEN A, KARHUNEN J, OJA E. Independent Component Analysis[M]. CanadaJohnWiley and Sons,2001.
[60] TONG L, SOON V C, HUANG Y F, et al. AMUSE: A New Blind Identification Algo-rithm[C]. IEEE International Symposium on Circuits and Systems.1990:1784–1787.
[61] TONG L, WEN LIU R, SOON V C, et al. Indeterminacy and Identifiability of Blind Identi-fication[J]. IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS,1991,38:499–509.
[62] BROWNE M, CUTMORE T. Low-probability event-detection and separation via statisticalwavelet thresholding: an application to psychophysiological denoising[J]. Clinical Neuro-physiology,2002,113:1403–1411.
[63] HE P, WILSON G, RUSSELL C. Removal of ocular artifacts from electroencephalogram byadaptive filtering[J]. Med Biol Eng Comput,2004,42:407–412.
[64] CICHOCKI A, SHISHKIN S L, MUSHA T, et al. EEG filtering based on blind source separa-tion (BSS) for early detection of Alzheimer’s disease[J]. Clinical Neurophysiology,2005,116:729–737.
[65] BELL A J, SEJNOWSKI T J. An information-maximisation approach to blind separationand blind deconvolution[J]. Neural Computation,1995,7:1004–1034.
[66] HYVARINEN A, OJA E. Independent Component Analysis: Algorithms and Applica-tions[J]. Neural Networks,2000,13:411–430.
[67] YANG H H, AMARI S. Adaptive on-line learning algorithms for blind separation-max-imum entropy and minimum mutural information[J]. Neural Computation,1997,9:1457–1482.
[68] PFURTSCHELLER G, NEUPER C, FLOTZINGER D, et al. EEG-based discrimination be-tween imagination of right and left hand movement[J]. Electroencephalography and ClinicalNeurophysiology,1997,103:642–651.
[69] KOLES Z. The quantitative extraction and topographic mapping of the abnormal compo-nents in the clinical EEG[J]. Electroencephalography and Clinical Neurophysiology,1997,79:440–447.
[70] PARRA L C, SPENCE C D, GERSON A D, et al. Recipes for the linear analysis of EEG[J].NeuroImage,2005,28:326–31.
[71] ANDERSON C W, STOLZ E A, SHAMSUNDER S. Multivariate Autoregressive Modelsfor Classification of Spontaneous Electroencephalogram During Mental Tasks[J]. IEEETransactions on Biomediacal Engineering,1998,45:277–286.
[72] PRESTLEY M B. Spectral Analysis and Time Series[M]. Academic Press,1981.
[73] MORMANN F, LEHNERTZ K, DAVID P, et al. Mean phase coherence as a measure for phasesynchronization and its application to the EEG of epilepsy patients[J]. Physica D: NonlinearPhenomena,2000,144:358–369.
[74] RICHMAN J S, MOORMAN J R. Physiological time-series analysis using approximate en-tropy and sample entropy[J]. American Journal of Physiology-Heart and Circulatory Phys-iology,2000,278:H2039–H2049.
[75] MASSEY F. The Kolmogorov-Smirnov Test for Goodness of Fit[J]. Journal of the AmericanStatistical Association,1951,46:68–78.
[76] BISHOP C. Pattern Rcognition and Machine Learning[M]. SingaporeSpringer,2006.
[77] SHUMWAY R, STOFFER D. Time Series Analysis and Its Applications: With R Exam-ples[M].2nd. Springer,2006.
[78] GHAHRAMANI Z, HINTON G E. Parameter estimation for linear dynamical systems[R].Technical Report CRG-TR-96-2, University of Toronto,1998.
[79] TORRE F D L, BLACK M J. A Framework for Robust Subspace Learning[J]. Interna-tional Journal of Computer Vision,2003,54:117–142.
[80] CAMPBELL N. Robust procedures in multivariate analysis I: Robust covariance estima-tion[J]. Applied Statistics,1980,29:231–237.
[81] HOLLAND P W, WELSCH R E. Robust regression using iteratively reweighted leastsquares[J]. Communications in Statistics-Theory and Methods,1977,6:813–827.
[82] XU L, YULLE A L. Robust principal component analysis by self-organizing rules based onstatistical physics approach[J]. IEEE Transactions on Neural Networks,1995,6:131–143.
[83] GABRIEL K, ODOROFF C. Resistant lower rank approximation of matrices[J]. Data Anal-ysis and Informatics, III,1984:23–30.
[84] GREENACRE M. Correspondence analysis of multivariate categorical data by weighted leastsquares[J]. Biometrika,1988,75:457–467.
[85] BLACK M, RANGARAJAN A. On the unification of line processes, outlier rejection, and ro-bust statistics with applications in early vision[J]. International Journal of Computer Vision,1996,25:57–92.
[86] KE Q, KANADE T. Robust L1Norm Factorization in the Presence of Outliers and MissingData by Alternative Convex Programming[C]. IEEE Conf. Computer Vision and PatternRecognition.2004:592–599.
[87] DING C, ZHOU D, HE X, et al. R1-PCA: Rotational Invariant L1-norm Principal Compo-nent Analysis for Robust Subspace Factorization[C]. International Conference on MachineLearning.2006:281–288.
[88] KWAK N. Principal Component Analysis based on L1-norm Maximization[J]. IEEE Trans-actions on Pattern Analysis and Machine Intelligence,2008,30:1672–1680.
[89] ZOU H, HASTIE T, TIBSHIRANI R. Sparse principal component analysis[J]. Journal ofComputational and Graphical Statistics,2006,15:265–286.
[90] ZOU H, HASTIE T. Regularization and variable selection via the elastic net[J]. Journal ofthe Royal Statistical Society, Series B,2005,67:301–320.
[91] EFRON B, HASTIE T, JOHNSTONE I, et al. LEAST ANGLE REGRESSION[J]. The Annalsof Statistics,2004,32:407–499.
[92] PENG H, LONG F, DING C. Feature Selection Based on Mutual Information: Criteria ofMax-Dependency, Max-Relevance, and Min-Redundancy[J]. IEEE Transactions on PatternAnalysis and Machine Intelligence,2005,27:1226–1238.
[93] SHI L, LU B. EEG-based vigilance estimation using extreme learning machines[J]. Neuro-computing(2012),http://dx.doi.org/10.1016/j.neucom.2012.02.041.
[94] HUANG G B, ZHU Q Y, SIEW C K. Extreme Learning Machine: Theory and Applica-tions[J]. Neurocomputing,2006,70:489–501.
[95] HUANG G B, CHEN L, SIEW C K. Universal Approximation Using Incremental Construc-tive Feedforward Networks with Random Hidden Nodes[J]. IEEE Transactions on NeuralNetworks,2006,17:879–892.
[96] HUANG G B, CHEN L. Convex Incremental Extreme Learning Machine[J]. Neurocomput-ing,2007,70:3056–3062.
[97] HUANG G B, CHEN L. Enhanced Random Search Based Incremental Extreme LearningMachine[J]. Neurocomputing,2008,71:3460–3468.
[98] VAPNIK V, GOLOWICH S E, SMOLA A J. Support Vector Method for Function Approxi-mation, Regression Estimation, and Signal Processing[J]. Advances in Neural InformationProcessing Systems,1996,9:281–287.
[99] SMOLA A J, SCHOLKOPF B. A tutorial on support vector regression[J]. Statistics andComputing,2004,14:199–222.
[100] SHI J, MALIK J. Normalized Cuts and Image Segmentation[J]. IEEE Transactions onPattern Analysis and Machine Intelligence,22:888–905.
[101] YU S, SHI J. Multiclass spectral clustering[C]. Proc. Int. Conf. on computer vision.2003:313–319.
[102] MAKEIG S, JUNG T. Changes in alertness are a principal component of variance in tehEEG spectrum[J]. NeuroReport,1995,7:213–216.
[103] TENENBAUM J, DE SILVA V, LANGFORD J. A global geometric framework for nonlineardimensionality reduction[J]. Science,2000,290:2319–2323.
[104] DE SILVA V, TENENBAUM J. Global versus Local Methods in Nonlinear DimensionalityReduction[C]. Advances in Neural Information Processing Systems.2003:705–712.
[105] CHANG C C, LIN C J. LIBSVM: A Library for Support Vector Machines[R]. http://www.csie.ntu.edu.tw/cjlin/libsvm,2001.
[2] DAWSON D, LAMOND N, DONKIN K. Quantitative similarity between the cognitive psy-chomotor performance decrement associated with sustained wakefulness and alcohol intoxi-cation[C]. Proceedings of the Third International Conference on Fatigue and Transportation.1998:44–45.
[3]陈伟.疲劳驾驶:法国人的防范之策[J].社会发展,2000,10:43–44.
[4] KENNETH P, WRIGHT J, JOSEPH H. Relationship between alertness,performance,and bodytemperature in humans[J]. American Journal of Physiology,2002,283(6):R1370–R1377.
[5] CAJOCHEN C, KHALSA S, WYATT J. EEG and ocular correlates of circadian melatoninphase and human performance decrements during sleep loss[J]. American Journal of Phys-iology,1999,277(3):R640–R649.
[6] DINGES D F, MALLIS M, MAISLIN G, et al. Evaluation of techniques for ocular measure-ment as an index of fatigue and the basis for alertness management[R]. National HighwayTraffic Safety Administration, USA,1998.
[7] JI Q, YANG X. Real-time eye, gaze, and face pose tracking for monitoring driver vigi-lance[J]. Real-Time Imaging,2002,8(5):357–377.
[8] BOVERIS S. Driver fatigue monitoring technologies and future ideas[C]. Proceedings ofAWAKE Road Safety Workshop. Balocco, ItalyElsevier Science Ltd,2004.
[9] MATTHEWS R, MCDONALD N J, HERVIEUX P, et al. A wearable physiological sensorsuite for unobtrusive monitoring of physiological and cognitive state[C]. Proc. IEEE Int.Conf. of Engineering in Medicine and Biology Society.2007:5276–5281.
[10] WANG X W, SHI L C, LU B L. A Survey on Technology of Dry Electrodes for EEGRecording[J]. Chinese Journal of Biomedical Engineering,2010,29:777–784.
[11] PFURTSCHELLER G, ARANIBAR A. Event-related cortical desynchronization detected bypower measurements of scalp EEG[J]. Electroencephalography Clinical Neurophysiology,1977,42(6):817–826.
[12] MATOUSEK M, PETERSEN I. A method for assessing alertness fluctuations from EEGspectra[J]. Electroencephalography Clinical Neurophysiology,1983,55(1):108–113.
[13]李文磊.基于DSP的疲劳驾驶实时监测系统研究[D].南京:南京理工大学,2007.
[14] HYOKI K, SHIGETA M, TSUNOB N. Quantitative electro-oculography and electro-en-cephalography as indices of alertness[J]. Electroencephalography and clinical Neurophysi-ology,1998,106(3):213–219.
[15] MAGOSSO E, PROVINI F, MONTAGNA P. A wavelet based method for automatic detec-tion of slow eye movements: A pilot study[J]. Medical Engineering and Physics,2006,28(9):860–875.
[16] MAGOSSO E, URSINO M, ZANIBONI A. Visual and computer-based detection of sloweye movements in overnight and24-h EOG recordings[J]. Clinical Neurophysiology,2007,118(5):1122–1133.
[17] ATIENZA M, CANTERO J L, STICKGOLD R. Eyelid movements measured by Nightcappredict slow eye movements during quiet wakefulness in humans[J]. Journal of Sleep Re-search,2004,13(1):25–29.
[18] KAIDA K, AKERSTED T, KECKLUND G. Use of Subjective and Physiological Indicatorsof Sleepiness to Predict Performance during a Vigilance Task[J]. Industrial Health,2007,45(4):520–526.
[19] JAMMES B, SHARABTY H, ESTEVE D. Automatic EOG analysis: A first step toward auto-matic drowsiness scoring during wake-sleep transitions[J]. Somnologie,2008,12(3):227–232.
[20] OKEN B, SALINSKY M, ELSAS S. Vigilance, alertness, or sustained attention: physiologi-cal basis and measurement[J]. Clinical Neurophysiology,2006,117:1885–1901.
[21]毛喆.基于驾驶员生理特征分析的驾驶疲劳状态识别方法研究[D].武汉:武汉理工大学,2006.
[22] KUBICKI S, HERRMANN W, FICHTE K, et al. Reflections on the topics: EEG frequencybands and regulation of vigilance[J]. Pharmakopsychiatr Neuropsychopharmakol,1979,12(2):237–245.
[23] MAKEIG S, INLOW M. Lapses in alertness: coherence of fluctuations in performance andEEG spectrum[J]. Electroencephalography and Clinical Neurophysiology,1993,86(1):23–35.
[24] MAKEIG S. Using Feedforward Neural Networks to Monitor Alertness from Changes inEEG Correlation and Coherence[J]. Advances in Neural Information Processing Systems,1996,8:931–937.
[25] T.P.JUNG, MAKEIG S, STENSMO M, et al. Estimating Alertness from the EEG PowerSpectrum[J]. IEEE Transactions on Biomedical Engineering,1997,44(1):60–69.
[26] HUANG R, JUNG T, DELORME A, et al. Tonic and phasic electroencephalographic dynam-ics during continuous compensatory tracking[J]. NeuroImage,2008,39(4):1896–1909.
[27] LIN C, WU R, LIANG S, et al. EEG-Based Drowsiness Estimation for Safety Driving UsingIndependent Component Analysis[J]. IEEE Transaction on Circuits and Systems-I,2005,52(12):2726–2738.
[28] LIN C, KO L, CHUNG I, et al. Adaptive EEG-Based Alertness Estimation System by UsingICA-Based Fuzzy Neural Networks[J]. IEEE Transactions on Circuits and Systems-I,2006,53(11):2469–2476.
[29] LIN C, HUANG K, CHAO C, et al. Tonic and phasic EEG and behavioral changes inducedby arousing feedback[J]. NeuroImage,2010,52(2):633–642.
[30] LIN C, KO L, CHIOU J, et al. Noninvasive neural prostheses using mobile and wirelessEEG[J]. Proc. IEEE,2008,96(7):1167–1183.
[31] CHI Y M, CAUWENBERGHS G. Micropower Non-contact EEG Electrode with ActiveCommon-Mode Noise Suppression and Input Capacitance Cancellation[J]. Proc. IEEE Int.Conf. of Engineering in Medicine and Biology Society,2009:4218–4221.
[32] DAVIDSON P R, JONES R D, PEIRIS M T R. EEG-Based Lapse Detection With HighTemporal Resolution[J]. IEEE Transactions on Biomedical Engineering,2007,54(5):832–839.
[33]赵仑. ERP实验教程[M].天津:天津社会科学院出版社,2004:11–12.
[34] NORDBY H, HUGDAHL K, STICKGOLD R, et al. Event-related potentials (ERPs) to deviantauditory stimuli during sleep and waking[J]. NeuroReport,1996,7:1082–1086.
[35]傅佳伟,石立臣,吕宝粮.基于EEG的警觉度分析与估计研究综述[J].中国生物医学工程学报,2009,28(4):589–596.
[36] RICHMAN J S, MOORMAN J R. Physiological time-series analysis using approximate en-tropy and sample entropy[J]. American Journal of Physiology,2000,278(6):H2039–H2049.
[37] ESTELLER R, VACHTSEVANOS G, ECHAUZ J, et al. A comparison of waveform fractal di-mension algorithms[J]. IEEE Transactions on Circuits and Systems I: Fundamental Theoryand Applications,2001,48(2):177–183.
[38] PARDEY J, ROBERTS S, TARASSENKO L. A review of parametric modelling techniquesfor EEG analysis[J]. Medical Engineering&Physics,1996,18(1):2–11.
[39] FATOURECHI M, BASHASHATI A, WARD R K, et al. EMG and EOG artifacts in braincomputer interface systems: A survey[J]. Clinical Neurophysiology,2007,118:480–494.
[40] GRATTON G. Dealing with artifacts: the EOG contamination of the event-related brainpotential[J]. Behav Res Methods, Instrum, Comput,1998,30:44–53.
[41] MILLAN J, FRANZE M, MOURINO J, et al. Relevant EEG features for the classi?cation ofspontaneous motor-related tasks[J]. Biol Cybern,2002,86:89–95.
[42] GOTMAN J, SKUCE D, THOMPSON C, et al. Clinical applications of spectral analysis andextraction of features from electroencephalograms with slow waves in adult patients[J]. BiolCybern,1973,35:225–235.
[43] CROFT R, CHANDLER J, BARRY R, et al. EOG correction: a comparison of four meth-ods[J]. Psychophysiology,2005,42:16–24.
[44] LAGERLUND T, SHARBROUGH F, BUSACKER N. Spatial filtering of multichannel elec-troencephalographic recordings through principal component analysis by singular value de-composition[J]. Journal of clinical neurophysiology,1997,14:73–82.
[45] JUNG T, MAKEIG S, WESTERFIELD M, et al. Removal of eye activity artifacts from visualevent-related potentials in normal and clinical subjects[J]. Clinical Neurophysiology,2000,111:1745–1758.
[46] LACHAUX J, RODRIGUEZ E, MARTINERIE J, et al. Measuring Phase Synchrony in BrainSignals[J]. Human Brain Mapping,1999,8:194–208.
[47] PEIRIS M, JONES R, DAVIDSON P, et al. Identification of Vigilance Lapses usingEEG/EOG by Expert Human Raters[C]. Proc. IEEE Int. Conf. of Engineering in Medicineand Biology Society.2005:5735–5737.
[48] JOHNS M, TUCKER A, CHAPMAN R. A New Method for Monitoring the Drowsiness ofDrivers[C]. Proc. Int. Conf. on Fatigue Management in Transportation Operations.2005:11–15.
[49] GONCHAROVA I I, MCFARLAND D J, VAUGHAN T M, et al. EMG contamination of EEG:spectral and topographical characteristics[J]. Clinical Neurophysiology,2003,114:1580–1593.
[50] BLUME W T, KAIBARA M, YOUNG G B. Atlas of Adult Electroencephalography[M].Lippincott Williams&Wilkins,2002.
[51] BARLOW J S. Artifact processing (rejection and minimization) in EEG data processing[J].Clinical Neurophysiology,1986,2:15–62.
[52] MA J X, SHI L C, LIANG LU B. Vigilance Estimation by Using ElectrooculographicFeatures[C].32nd Annual International Conference of the IEEE Engineering in Medicineand Biology Society (EMBC10).2010:6591–6594.
[53] CAI H Y, MA J X, SHI L C, et al. A novel method for EOG features extraction from theforehead[C].33rd Annual International Conference of the IEEE Engineering in Medicineand Biology Society (EMBC11).2011:3075–3078.
[54] ANDERER P, ROBERTS S, SCHLOGL A, et al. Artifact Processing in Computerized Analy-sis of Sleep EEG-A Review[J]. Neuropsychobiology,1999,40:150–157.
[55] OCHOA C J, POLICH J. P300and blink instructions[J]. Clinical Neurophysiology,2000,111:93–98.
[56] VAN DE VELDE M, VAN ERP G, CLUITMANS P J. Detection of muscle artefact in the nor-mal human awake EEG[J]. Electroencephalography and Clinical Neurophysiology,1998,107:149–158.
[57] VAN DE VELDE M, VAN ERP G, CLUITMANS P J. Context related artefact detection inprolonged EEG recordings[J]. Computer Methods and Programs in Biomedicine,1999,60:183–196.
[58] CHOI S, CICHOCKI A, PARK H M, et al. Blind source separation and independent compo-nent analysis: a review[J]. Neural Inf Process-Lett Rev,2005,6:1–57.
[59] HYVARINEN A, KARHUNEN J, OJA E. Independent Component Analysis[M]. CanadaJohnWiley and Sons,2001.
[60] TONG L, SOON V C, HUANG Y F, et al. AMUSE: A New Blind Identification Algo-rithm[C]. IEEE International Symposium on Circuits and Systems.1990:1784–1787.
[61] TONG L, WEN LIU R, SOON V C, et al. Indeterminacy and Identifiability of Blind Identi-fication[J]. IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS,1991,38:499–509.
[62] BROWNE M, CUTMORE T. Low-probability event-detection and separation via statisticalwavelet thresholding: an application to psychophysiological denoising[J]. Clinical Neuro-physiology,2002,113:1403–1411.
[63] HE P, WILSON G, RUSSELL C. Removal of ocular artifacts from electroencephalogram byadaptive filtering[J]. Med Biol Eng Comput,2004,42:407–412.
[64] CICHOCKI A, SHISHKIN S L, MUSHA T, et al. EEG filtering based on blind source separa-tion (BSS) for early detection of Alzheimer’s disease[J]. Clinical Neurophysiology,2005,116:729–737.
[65] BELL A J, SEJNOWSKI T J. An information-maximisation approach to blind separationand blind deconvolution[J]. Neural Computation,1995,7:1004–1034.
[66] HYVARINEN A, OJA E. Independent Component Analysis: Algorithms and Applica-tions[J]. Neural Networks,2000,13:411–430.
[67] YANG H H, AMARI S. Adaptive on-line learning algorithms for blind separation-max-imum entropy and minimum mutural information[J]. Neural Computation,1997,9:1457–1482.
[68] PFURTSCHELLER G, NEUPER C, FLOTZINGER D, et al. EEG-based discrimination be-tween imagination of right and left hand movement[J]. Electroencephalography and ClinicalNeurophysiology,1997,103:642–651.
[69] KOLES Z. The quantitative extraction and topographic mapping of the abnormal compo-nents in the clinical EEG[J]. Electroencephalography and Clinical Neurophysiology,1997,79:440–447.
[70] PARRA L C, SPENCE C D, GERSON A D, et al. Recipes for the linear analysis of EEG[J].NeuroImage,2005,28:326–31.
[71] ANDERSON C W, STOLZ E A, SHAMSUNDER S. Multivariate Autoregressive Modelsfor Classification of Spontaneous Electroencephalogram During Mental Tasks[J]. IEEETransactions on Biomediacal Engineering,1998,45:277–286.
[72] PRESTLEY M B. Spectral Analysis and Time Series[M]. Academic Press,1981.
[73] MORMANN F, LEHNERTZ K, DAVID P, et al. Mean phase coherence as a measure for phasesynchronization and its application to the EEG of epilepsy patients[J]. Physica D: NonlinearPhenomena,2000,144:358–369.
[74] RICHMAN J S, MOORMAN J R. Physiological time-series analysis using approximate en-tropy and sample entropy[J]. American Journal of Physiology-Heart and Circulatory Phys-iology,2000,278:H2039–H2049.
[75] MASSEY F. The Kolmogorov-Smirnov Test for Goodness of Fit[J]. Journal of the AmericanStatistical Association,1951,46:68–78.
[76] BISHOP C. Pattern Rcognition and Machine Learning[M]. SingaporeSpringer,2006.
[77] SHUMWAY R, STOFFER D. Time Series Analysis and Its Applications: With R Exam-ples[M].2nd. Springer,2006.
[78] GHAHRAMANI Z, HINTON G E. Parameter estimation for linear dynamical systems[R].Technical Report CRG-TR-96-2, University of Toronto,1998.
[79] TORRE F D L, BLACK M J. A Framework for Robust Subspace Learning[J]. Interna-tional Journal of Computer Vision,2003,54:117–142.
[80] CAMPBELL N. Robust procedures in multivariate analysis I: Robust covariance estima-tion[J]. Applied Statistics,1980,29:231–237.
[81] HOLLAND P W, WELSCH R E. Robust regression using iteratively reweighted leastsquares[J]. Communications in Statistics-Theory and Methods,1977,6:813–827.
[82] XU L, YULLE A L. Robust principal component analysis by self-organizing rules based onstatistical physics approach[J]. IEEE Transactions on Neural Networks,1995,6:131–143.
[83] GABRIEL K, ODOROFF C. Resistant lower rank approximation of matrices[J]. Data Anal-ysis and Informatics, III,1984:23–30.
[84] GREENACRE M. Correspondence analysis of multivariate categorical data by weighted leastsquares[J]. Biometrika,1988,75:457–467.
[85] BLACK M, RANGARAJAN A. On the unification of line processes, outlier rejection, and ro-bust statistics with applications in early vision[J]. International Journal of Computer Vision,1996,25:57–92.
[86] KE Q, KANADE T. Robust L1Norm Factorization in the Presence of Outliers and MissingData by Alternative Convex Programming[C]. IEEE Conf. Computer Vision and PatternRecognition.2004:592–599.
[87] DING C, ZHOU D, HE X, et al. R1-PCA: Rotational Invariant L1-norm Principal Compo-nent Analysis for Robust Subspace Factorization[C]. International Conference on MachineLearning.2006:281–288.
[88] KWAK N. Principal Component Analysis based on L1-norm Maximization[J]. IEEE Trans-actions on Pattern Analysis and Machine Intelligence,2008,30:1672–1680.
[89] ZOU H, HASTIE T, TIBSHIRANI R. Sparse principal component analysis[J]. Journal ofComputational and Graphical Statistics,2006,15:265–286.
[90] ZOU H, HASTIE T. Regularization and variable selection via the elastic net[J]. Journal ofthe Royal Statistical Society, Series B,2005,67:301–320.
[91] EFRON B, HASTIE T, JOHNSTONE I, et al. LEAST ANGLE REGRESSION[J]. The Annalsof Statistics,2004,32:407–499.
[92] PENG H, LONG F, DING C. Feature Selection Based on Mutual Information: Criteria ofMax-Dependency, Max-Relevance, and Min-Redundancy[J]. IEEE Transactions on PatternAnalysis and Machine Intelligence,2005,27:1226–1238.
[93] SHI L, LU B. EEG-based vigilance estimation using extreme learning machines[J]. Neuro-computing(2012),http://dx.doi.org/10.1016/j.neucom.2012.02.041.
[94] HUANG G B, ZHU Q Y, SIEW C K. Extreme Learning Machine: Theory and Applica-tions[J]. Neurocomputing,2006,70:489–501.
[95] HUANG G B, CHEN L, SIEW C K. Universal Approximation Using Incremental Construc-tive Feedforward Networks with Random Hidden Nodes[J]. IEEE Transactions on NeuralNetworks,2006,17:879–892.
[96] HUANG G B, CHEN L. Convex Incremental Extreme Learning Machine[J]. Neurocomput-ing,2007,70:3056–3062.
[97] HUANG G B, CHEN L. Enhanced Random Search Based Incremental Extreme LearningMachine[J]. Neurocomputing,2008,71:3460–3468.
[98] VAPNIK V, GOLOWICH S E, SMOLA A J. Support Vector Method for Function Approxi-mation, Regression Estimation, and Signal Processing[J]. Advances in Neural InformationProcessing Systems,1996,9:281–287.
[99] SMOLA A J, SCHOLKOPF B. A tutorial on support vector regression[J]. Statistics andComputing,2004,14:199–222.
[100] SHI J, MALIK J. Normalized Cuts and Image Segmentation[J]. IEEE Transactions onPattern Analysis and Machine Intelligence,22:888–905.
[101] YU S, SHI J. Multiclass spectral clustering[C]. Proc. Int. Conf. on computer vision.2003:313–319.
[102] MAKEIG S, JUNG T. Changes in alertness are a principal component of variance in tehEEG spectrum[J]. NeuroReport,1995,7:213–216.
[103] TENENBAUM J, DE SILVA V, LANGFORD J. A global geometric framework for nonlineardimensionality reduction[J]. Science,2000,290:2319–2323.
[104] DE SILVA V, TENENBAUM J. Global versus Local Methods in Nonlinear DimensionalityReduction[C]. Advances in Neural Information Processing Systems.2003:705–712.
[105] CHANG C C, LIN C J. LIBSVM: A Library for Support Vector Machines[R]. http://www.csie.ntu.edu.tw/cjlin/libsvm,2001.