摘要
随着微电子技术的不断发展，电子元件朝着超薄型、高密度、细间距、无铅化方向快速发展，对印刷电路板的焊接质量检测提出了更高要求，传统上依靠人工目检对印刷电路板焊点质量进行检测的方式很难满足生产要求。而自动光学检测系统(Automatic OpticalInspection, AOI)能检测到微小组件和较高安装密度的PCB（Printed Circuit Board,PCB），并且也有的统一的检测标准、快速度、利于产品质量管理等特点，但是现有的AOI系统在实际应用中存在以下几个方面的问题：编程较为复杂，特别是调试程序的时间过长； AOI系统检测存在漏判和误判的问题；作为在线检测系统，AOI的检测速度有待提高。因此，本文分别在定位、编程和缺陷检测这三个方面对AOI进行优化及改进研究。主要研究内容如下：
（1）提出一种新的基于三层MARK点位置定位方法。第一层是全局对齐，补偿PCB加载过程位置误差；第二层是单板对齐，补偿的单板装配的位置误差,最后一层是局部视场(Field Of View,FOV)对齐，补偿的局部凹凸不平的表面定位误差。实验结果证明了提出的方法改善了位置定位的精度。
自动光学检测的路径规划问题可以建模为一个标准的旅行商问题(TravellingSalesman Problem,TSP),为最小化总工作时间，提出了一种基于Hopfield网络新方法的路径规划问题。通过Hopfield网络获得了一个优化路径。
（2）提出了一种基于像素统计建模原理的AOI的测试算法，这种算法运用的是一种统计建模方法，通过来对焊点样本图进行灰度级别的建模来达成目标。该方法操作过程的第一步是建立一个合格样本的模板图，接下来通过两个步骤的定位方法，运用对待测的元件图及与模板图配对比较，计算两者差值的原理，来判定其是否达标。实验结果显示,基于统计建模的图像配对算法使得误报率和漏报率都达到实际使用需求；另外，还有效地缩短了编程和调试时间,降低了人为因素对检测结果的影响,提高了AOI算法的效率和准确率。近年来,该算法已经被广泛应用于实际生产过程中,给用户带来了极便捷的使用体验。
（3）由于在焊点检测过程中，漏报的损失要远大于误报的损失，因此本文提出了一个新的基于最小风险贝叶斯分类器来进行焊点合格/缺陷的二分方法。Adaboost的作为一个强分类器被引进。实验结果表明,最小风险贝叶斯分类器在二分焊点的过程中有一个良好的性能、Adaboost算法检验结果也进一步验证了最小风险贝叶斯分类器的良好性能。
（4）提出一种神经网络结合遗传算法的诊断焊点缺陷的方法。提取焊点图像的14个特征作为分类器的输入。神经网络很容易变得过度学习,因为这些输入特征不是相互独立的缘故,因此遗传算法被引入来选择和去除冗余的输入特征。实验结果证明,神经网络结合遗传算法减少了输入特征的数量并取得一个令人满意的识别率。
（5）提出一种基于信息增益的特征选择结合对焊点分类的两级分类器框架。该方法首先获得的焊点图像。然后提取颜色特征（平均灰度和高亮像素比例）和模板匹配特征。经基于信息增益特征选择后,用贝叶斯算法,每个焊点被分为合格/不合格两类。如果焊点判为不合格,焊点将被支持向量机(Support Vector Machine)分为一个预定义多分类类型,第二阶段分类器选择是基于各种分类器分类性能评估的。实验结果表明,该方案不仅更高效,也提高了识别率,因为它减少了提取特征数量的需要。
最后，在总结全文内容的基础上，本文还对未来AOI研究方向进行了展望。
With the development of the microelectronics technology, electronic components rapidalydevelop towards ultra thin, high density, small spacing and lead-free.This makes a higherrequirement to the welding quality of printed circuit board, visual inspection of solder joint onthe printed circuit board is difficult to meet the production requirements. And Automatic OpticalInspection (AOI) system can inspection high density of micro components pasted on the PCB(Printed Circuit Board, PCB), and the this Inspection method is standard unified, fast speed,online management to product quality, but the existing AOI system has the following severalaspects of the problems in the practical application, the programming is relatively complex,especially the debugging time is too long; AOI system has high faults missed and false alarm; Asan online examination system, the AOI test speed still needs to improve. In this disserrtation,therefore, study on the three aspects of optimization and improvement AOI: positioning,programming, and defect detection. Main research contents are as follows:
(1) A new position method based on three layers of MARK point is developed. the first layeris the whole board alignment which compensate the position error of the PCB loading process,the second layer is the single board alignment which compensate the position error of the singleboard assembly, the last layer is the partial FOV(field of view) alignment which compensate theposition error of the partial uneven surface. The experiment result proved the proposed positionmethod improved the position accuracy obviously.
The path planning problem of AOI can be modeled as a standard travelling salesman problem(TSP), to minimize the overall working time, a new method based on Hopfield net algorithm isproposed to optimize the path planning problem. The experiments show the Hopfield netoptimize the planning path.
(2) A new method is presented of image matching based on statistic modeling. a standardtemplate image was formed through statistic the gray value of good samples. The testingcomponent image and the model image were compared after alignment, then the comparisionresult determines whether the testing component image qualified or not.The experimental resultsshow that: the proposed method meets practical application requirementsin the rate of false alarmand faults missed And effectively simplifies the programming and debugging, reduces thehuman factors that influence the result of the test, improved the AOI accuracy and efficiency. At present, the algorithm has been successfully applied in actual production, to give users a greatconvenience and benefits.
(3) In the process of solder joint inspection, the loss of faults missed is greater than the loss offalse alarm, so this paper proposes a new method based on the minimum risk bayes classifier forsolder joint qualified/defect detection. Adaboost was introduced as a strong classifier.Experimental show that the minimum risk bayes classifier in the process of classfying solderjoint has the best inspection result, the inspection result of Adaboost algorithm is furthervalidated the good performance of minimum risk bayes classifier.
(4) A neural network combined with genetic algorithm for the diagnosis of solder joint ispresented.14features are extracted as input features for the classification. The neural network iseasily become over-fitting because these input features are not independent of each other, so thegenetic algorithm is introduced to select and remove redundant input features. The experimentalresults have proved that the neural network combined with genetic algorithm reduced the numberof input feature and had a satisfying recognition rate.
(5) A feature selection and a two-stage classifier for solder joint inspection have beenproposed. The images of solder joint can be obtained. The color features including the averagegray level and the percentage of highlights and the template matching feature are extracted. Afterfeature selection of information gain, based on the algorithm of Bayes, each solder joint isclassified into by its qualification. If the solder joint fail in the qualification test, the solder jointis classified into one of pre-defined types based on the support vector machine (SVM), thechoice of the second stage classifier is based on the performance evaluation of various classifiers.The experimental results showed that the proposed scheme not only more efficient, but alsoincreasing the recognition rate, since it reduces the number of needed extracted features.
Finally, the future research direction of the AOI is prospected based on the summarizationof the whole paper.
引文
[1]中国社会科学院工业经济研究所.2008中国工业发展报告—中国工业改革开放30年[M].北京:经济管理出版社,2008
[2]周德俭，吴兆华.组装工艺技术[M].北京:国防工业出版社,2002
[3]王亚雄.表面贴装技术(SMT)及应用[J].表面贴装技术,2003,9:21-25
[4]2008年1-11月电子信息产业主营业务同比增长16%[N].中国电子报,2009年
[5]鲜飞. SMT测试技术介绍[J].测量与分析,2003增刊:11-12
[6] Niereyer T.印刷电路板生产用自动光学检测系统[J].世界产品与技术,2002,(10):55-57
[7] Raymond D W, Haigh D F. Why automate optical inspection?[A]. International TestConference [C].1997:1033
[8] Lu S.L, Zhang X.M, Kuang Y.C. Optimized design of an AOI illuminator [A].Proceedingsof2007Wavelet Analysis and Pattern Recognition[C]. HONGKONG,2007:924-928
[9] Elliott D.J. Integrated circuit fabrication technology [M]. New York: McGraw-Hill,1982
[10] Hams K.L, et al. Automated wafer inspection characterization [A]. Proc. SPIE Conf. onOptical Microlithography IV, Technology for the Next Decade[C]. Santa Clara, CA,1985:138-146
[11] Hams K.L, Sandland P. Singleton R. Wafer inspection automation: Current and futureneeds [J]. Solid State Technology,1983,(8):199-205
[12] Kayaalp J.R. A knowledge based automatic on-line wafer (IC) inspection system [A].VISION85Conference Proceedings [C]. Detroit, MI USA,1985:117-130
[13] Bolhouse V.C. Machine vision applications for high volume electronics manufacturing[A]. VISION85Conference Proceedings[C]. Detroit, MI USA,1985:75-98
[14] Blaustein P., Hahn S. Realtime inspection of wafer surfaces [J]. Solid State Technology,1989,32(12):27-29
[15] Marr D. W.H Freeman. Vision [M]. San Francisco.1982
[16]阮求琦.数字图像处理[M].北京:电子工业出版社,2003
[17] Darn E. The P300: A system for automatic patterned wafer inspection [J]. Machine Visionand Applications,1988:205-221
[18] Konishi T, Misono M, Kato T. New technique for inspecting charge-coupled device (CCD)wafers for defects [A]. SPIE Conf: on Robot Vision[C].1982:128-132
[19] Yoda, Y.O, Taniguchi Y., Ejiri M. An automatic wafer inspection machine using pipelinedimage processing techniques [J]. IEEE Trans. Pattern Analysis and Machine Intelligence,1988,1:4-16
[20]胡跃明,
[21]谭颖.自动光学检测在中国的应用现状和发展[J].微计算机信息,2006,(22):143-146
[22]谢宏威.印刷电路板焊点智能检测算法的研究[D].广州:华南理工大学,2011
[23] Wu Fupei, Zhang Xianmin and Kuang Yongcong.An AOI algorithm for PCB based onfeature extraction Inspection [A].7th World Congress on Intelligent Control andautomation[C].2008:240-247
[24]吴福培.炉后印刷电路板无铅焊点的自动光学检测理论与方法研究[D].广州:华南理工大学,2009
[25] Bai Chang-bing, Qi Chun. Fast detection of circular pcb mark using Hough transform [J].Opto-electronic Engineering,2005,32（9）:75-78
[26] Li Cheng-peng, Fan Yan-bin.3recognition methods and analysis of PCB Mark pointbased on HALCON[J].Journal of Foshan University (Natural Science Edition),2010,28（2）:29-33
[27] OuYang Gao-fei, Kuang Yong-cong. Accurate Image Capture for Surface MountingComponent Based on Subpixel Location [J].Opto-electronic engineering,2010,37（6）:16-22
[28] Park Tae-Hyoung, Kim Hwa-Jung. Path planning of automatic optical inspectionmachines for PCB assembly systems [A].2005IEEE International Symposium onComputational Intelligence in Robotics and Automation [C].2005:249-254
[29] Luo Bing and Zhang Yun. SMT automatic optical inspection path planning based on antcolony algorithm [J]. Chinese Journal of Scientific Instrument,2006,27(6):94-97
[30] Hopfield. J. Jand Tank.D.W.'Neural' Computation of Decisions in Optimization Problems[J]. Biological Cybernetics,1985,52(3):141-152
[31] Michael.T.G, Joseph S.B.M, Mark W.O. Approximate geometric pattern matching underrigid motions [J]. IEEE Transactions on Pattern Analysis and Machine Intelligence,1999,21(4):371-379
[32]何东健.数字图像处理[M].西安:西安电子科技大学出版社,2003
[33]李弼程,彭天强,彭波,等.智能图像处理技术[M].北京:电子工业出版社,2004
[34] David A, Forsyth J.P. Computer vision: a modern approach [M]. Beijing: PublishingHouse of Electronics Industry,2004
[35]冈萨雷斯伍兹著,阮求琦等译.数字图像处理[M].第二版.北京:电子工业出版社,2003
[36]季秀霞,季秀兰.基于无铅焊接的AOI技术[J].中国制造业信息化,2006,35(9):61-64
[37]王锦峰,严仕兴,统计建模技术在AOI中的应用[EB/OL].http//www.newsmt.com/xueyuan/pxshow? aspnewsid=2056
[38] Cyber Optics Corporation Inc. Flex automated optical inspection[EB/OL].www.cyberoptics.com,2007
[39] Liu H, Motoda H. Feature selection for knowledge discovery and data mining [M].Boston: Kluwer Academic Publishers,1998
[40] Molina L C, Llu′s Belanche, A`ngela Nebot. Feature selection algorithms: A survey andexperimental evaluation[R]. Barcelona: Universitat Politecnicade Catalunya,2002
[41] Holland J H. Adaptation in natural and artificial systems [M].Ann Arbor: University ofMichigan Press,1975
[42]葛继科,邱玉辉,吴春明,等.遗传算法研究综述[J].计算机应用研究,2008,25(10):2911-2916
[43] Goldberg D E. Genetic algorithms in search, optimization and machine learning[M].Boston: Addison-Wesley Longman Press,1989
[44]张文修,梁怡.遗传算法的数学基础[M].西安:西安交通大学出版社,2000
[45]徐宗本,张讲社,郑亚林.计算智能中的仿生学:理论与算法[M].北京:科学出版社,2003
[46] Bartlett S L, Besl P J, Cole C L, et al. Automatic solder joint inspection[J].PatternAnalysis and Machine Intelligence, IEEE Transactions on,1988,10(1):31-43
[47]盛雪莲,王乾人工神经网络[J].科技广场,2008,5:70-73
[48]张军,詹志辉计算智能[M].北京:清华大学出版社,2009
[49] Hebb D.The Organization of Behavior:A Neuropsychological Theory[M].New York.JohnWiley,1949
[50] Widrow B.Generalization and information storage in networks of adaline neurons[J].Self-organizing Systems.1962:435-461
[51] Rumelhart D E, Hinton G E, Williams R J.Learning internal representation by erropropagation[A].Parallel Distributed Processing [C].1986:318-362
[52] Ackley D H, Littman M L.Interation between learning and evolution [A]. Proceedings ofthe Second Conference on Artificial Life[C].Addison Wesley Pub,1992:487-509
[53] Homaifar A, Guan S.Traing weights of neural networks by genetic algorithms and messygenetic algorithms[A].Proc.2nd IASTED Int.Symo.Expert Systems and NeuralNetworks[C].M H Hamza,Ed.Anaheim,CA:Acta,1990:74-77
[54] Kozza J R, Rice J P.Genetic generation of both the weights and architecture for neuralnetwork[A]. IJCNN91[C]. Seattle,1991:397-404
[55] Frean M.The upstart algorithm: A method for construction and training feedforwardneural networks [J].Neural Computation,1990,2(2):198-209
[56] Angeline P J, Sauders G M, Pollack J B. An evolutionary algorithm that constructsrecurrent neural networks [A].IEEE Trans.Neural Networks(IJCNN91)[C], Seattle1991:397-404
[57] Hancock P J B, Smith L S, Phillips W A. A biologically supported error-correctinglearning rule [A].Artificial Neural Networks-ICANN-91[C].Amsterdam,North-Holland,1991:531-536
[58] Yan W, Zhu Z, Hu R.Hybrid genetic/BP algorithm and its application for radar targetclassification[A].Proc.1997IEEE National Aerospace and Electonics Conf,NAECON[C].1997:981-984
[59] Lee S W.Off-line recognition of totally unconstrained handwritten numerals usingmultilayer cluster neural network [J].IEEE Trans.Pattern Anal.Machine Intell,1996(18):648-652
[60] Greenwood G W.Training partially recurrent neural networks using networks usingevolutionary strategies [J].IEEE Trans.Speech Audio Processing,1997(5):192-194
[61] Kim J H, Cho H S.Neural network-based inspection of solder joints using a circularillumination[J].Image and Vsion Computing,1995(13):479-490
[62] Vijay Sankaran,Brent Chartrand. Automated inspection of solder joints-a neural networkapproach[A].IEEE/CPMT int’l Electronics Manufacturing TechnologySymposium[C].1995:232-237
[63] Kim T H, Cho T H, Moon Y S, Park S H. Visual inspection system for the classification ofsolder joints [J].pattern recognition,1999(32):565-575
[64] Ko K W, Cho H S. Solder joints inspection using a neural network and fuzzy rule-basedclassification method [J].IEEE TRANSACTIONS ON ELECTRONICS PACKAGINGMANUFACTURING,2000(23):93-103
[65] Belbachir A N, Lera M, Fanni A. An automatic optical inspection system for the diagnosisof printed circuits based on neural networks [A]. IEEE/IAS[C].2005:680-684
[66] Acciani G, Brunetti G, Fornarelli G. A multiple neural network system to classify solderjoints on integrated circuits [J].International Journal of Computational IntelligenceResearch,2006(2):337-348
[67] Ong T Y, Samad M M Ratnan. Solder joint inspection with multi-angle imaging and anartificial neural network[J].Int J Adv Manuf Technol,2008(38)455-462
[68] Luo Bing, Zhang Yun, Yu Guangzhu, Zhou Xianshan. ANN Ensembles Based MachineVision Inspection for Solder Joints [A].IEEE International Conference On Control andAutomation[C].2007:3111-3115
[69] Lin S C, Chou C H, Su C H. A Development of Visual Inspection System for SurfaceMounted Devices on Printed Circuit Board[A].IEEE Industrial Electonics SocietyIECON[C].2007:2440-2445
[70] Matsushima M, Kawai N, Fujie H, Yasude K, Fujimoto K. Visual Inspection of SolderingJoints by Neural Network with Multi-angle View and Principal ComponentAnalysis[J].Service Robotics and Mechatronics,2010:329-334
[71] Vapnik V. The Nature of Statistical Learning Theory [M]. New York:Spring-Verlag,1995
[72] Vapnik V. Statistical Learning Theory [M]. Addison-Wesley,1998
[73] Erin.L.Allwein, Robert.E.Schapire, Yoram Singer.Reducing multiclass to binary:Aunifying approach for margin classifiers [J]. Journal of Machine Learning Research,2000:1:113–141
[74] William W.Cohen, Yoram Singer.A simple,fast,and effective rule learner [A]. Proceedingsof the Sixteenth National Conference on Artificial Intelligence[C]. John Wiley&SonsLtd,1999:335-342
[75] Valiant L. A theory of learnability[J]. Communication ACM,1984,27:1134-1142
[76] Duda R, Hart P. Pattern Classification and Scene Analysis[M].John Wiley&Sons,1973(Pattern classification[M],2001)
[77]陶卿.机器学习及其应用[M].北京:清华大学出版社,2006
[78] Cortes C, Vapnik V. Support vector networks [J]. Machine Learning,1995,20:273-297
[79] Mangasarian Q L. Arbitrary-norm separating plane [J]. Operation Research Letters,1999,24(1):15-23
[80] Demirz A, Bennett K, Shawe-Taylor J. Linear Programming boosting via columngeneration [J]. Machine Learning,2002,46(1):225-254
[81] Keerthi S S, Shevade S K, Bhattacharyya C. A fast iterative nearest point algorithm fromSVMs [J]. An application to one-class classification. IEEE Trans. On Pattern Analysis andmachine intelligence,2002,9(4):1184-1199
[82] Sch lkopf B, Smola A J. Learning with Kernels. Cambridge, MA: The MIT Press,2002
[83] Sch lkopf B, Platt J, Shawe-Taylor J, Smola A J, Williamson R C.Estmating the supportof a high-dimensional distribution[J]. Neural Computation,2002,13(7):1443-1471
[84] R tsch G. Robust boosting via convex optimization. PhD Thesis, University ofPotsdam,2001
[85] Blumer A, Ehrenfenfeucht A,Haussler D, Warmuth MK. Learnability and the Vapnik-Chervokenkis dimension[J]. Journal of the ACM,1989,36(4):929-965
[86] Yun.T.S, Sim.K.J and Kim.H.J. Support vector machine-based inspection of solder jointsusing circular illumination [J]. Electronics Letters,2000,36(11):949-951
[87] Otsu N. A threshold selection method from gray-level histograms [J]. Automatica,1975,11(285-296):23-27
[88] Wu Fupei, Zhang Xianmin. Robust positioning algorithm of lead-free solder joints basedon gray-level integration projection [J]. Journal of South China University of Technology,2009,37(9):98-102
[89] Carsten Steger, Makus Urich. machine vision algorithms and applications[M].杨少荣.北京:清华大学出版社.2009
[90] Wang Ling and Zheng Dazhong. Study on TSP and Optimization Based on HopfieldNeural Network [J]. Control and decision,1999,14(6):669-674
[91] Sun Shouyu, Zheng Junli. A modified algorithm and theoretical analysis for Hopfieldnetwork solving TSP [J]. Acta Electronica Sinica,1995,23(1):73-78
[92] Ouglas W R, Dominic F H. Why automated optical inspection [A].Proc of InternationalTest Conference[C].1997:1033-1045
[93]贺珍真,张宪民,邝泳聪.电子元件的特征建模与检查算法研究[J].机电工程技术,2008,37(6):58-61
[94]王锦峰,严仕兴.统计建模技术在AOI中的应用[EB/OL] http//www. newsmt.com/xueyuan/pxshow? aspnewsid=2056
[95]吴福培,张宪民,邝泳聪.SMT炉后晶体管类元件无铅焊点的自动光学检测算法[J].焊接学报,2009,30(7):5-8
[96] Lin Shih-Chieh, Chou Chih-Hsien, Su Chia-Hsin.A development of visual inspectionsystem for surface mounted devices on printed circuit board[A]. The33rd AnnualConference of the IEEE Industrial Electronics Society[C]. Taipei, Taiwan,2007:2440-2445
[97] Loh Hong-Hai, Lu Ming-Sing. Printed circuit board inspection using image analysis[J].IEEE Transaction on Industry Application,1999,35(2):426-432
[98]吴浩,张宪民,邝泳聪,欧阳高飞，谢宏威.基于统计建模的电子元件焊点图像匹配算法[J].华南理工大学学报(自然科学版),2012,40(1):64-68
[99] Ong T Y, Samad Z, Ratnan M M. Solder joint inspection with multi-angle imaging and anartificial neural network [J]. International Journal Advance Manufaturing Technology,2008,(38):455-462
[100] Park J S, Tou J T. A solder joint insepction system for automated printed circuit boardmanufacturing [J]. IEEE International Conference on Robotics and Automation,1990(2):1290-1295
[101] Capson D W, Eng S K. A tiered color illumination approach for machine inspection ofsolder joint [J]. IEEE Transaction on Pattern Analysis and Machine Intelligence,1988,10(3):387-393
[102] Wu Fu-pei, Zhang Xian-min, Kuang Yong-cong.An AOI algorithm for PCB based onfeature extraction Inspection [J].7th World Congress on Intelligent Control andAutomation,2008:240-247
[103]吴福培,张宪民.基于模式匹配及其参数自适应的PCB焊点检测算法[J].光学精密工程,2009,17(10):2586-2593
[104] Wei Xu-bin, Wang Xiang-jun. Smoke target detection based on accumulated differentialimage sequence and gaussian background statistical model [J]. Journal of AstronauticMetrology and Measurement,2007,27(2):58-61
[105]卢盛林,张宪民.无铅焊点检测的光源分析与优化设计[J].光学精密工程，2008,16(8):1377-1383
[106]吴福培,张宪民,邝泳聪.无铅焊点鲁棒定位的灰度积分投影算法[J].华南理工大学学报:自然科学版,2009,37(9):99-102
[107]洪始良.印刷电路板自动给光学检测系统性能分析与优化[D].广州:华南理工大学,2010
[108] Bishop C M. Pattern recognition and machine learning [M]. New York: springer,2006
[109] Moganti M, Ercal F, Dagli C. H, and Tsunekawa S. Automatic PCB Inspection Algorithms:A Survey [J].Computer Vision and Image Understanding,1996,63:287-313
[110] Besl P J, Delp E J and Jain R.Automatic visual solder joint inspection [J]. IEEE J. RobAutomation,1985,1:42-56
[111] Bartlett S L, Besl P J and Jain R.Automatic solder joint inspection [J]. IEEE Trans.Pattern Anal. Mach. Intell,1988,10:31-43
[112] Driels M R, Nolan D J. Automatic defect classification of printed wiring board solderjoints [J]. IEEE Trans. Compon. Hybrids Manuf. Technol,1990,13:331-340
[113] Loh. H. Hand Lu.M. S. Printed circuit board inspection using image analysis [J]. IEEETransactions on Industry Applications,1999,35(2):426-432
[114] Chiu S and Perng M. Reflection-area-based feature descriptor for solder joint inspection[J] Machine Vision and Applications,2007,18(2):95-106
[115]姚旭,王晓丹,张玉玺,等.特征选择方法综述[J].控制与决策,2012,27(2)
[116]陈良龙.特征选择算法综述及基于weka的性能比较[EB/OL]http://wenku.baidu.com/view/2f7b18ece009581b6bd9eb46.html
[117] Gutierrez-Osuna R. Introduction to pattern analysis [J]. Lecture Notes, Texas A&MUniversity,2005
[118] Mitchell Tom. Machine Learning [M]. McGraw Hill,1997
[119] Holland. J. H. Adaptation in Natural and Artificial Systems [M]. Ann Arbor: Univ. ofMichigan Press,1975
[120] Fogel. D. B, System identification through simulated evolution: a machine learningapproach to modeling [M]. Neeham, Mass: Ginn press,1991
[121] Li K, Peng J X. System oriented neural networks—problem formulation, methodologyand application [J]. International Journal of Pattern Recognition and Artificial Intelligence,2006,20(02):143-158.
[122]研究者July.从决策树学习谈到贝叶斯分类算法、 EM、 HMM [EB/OL]http://blog.csdn.net/v_july_v/article/details/7577684
[123]张军.计算智能[M].北京：清华大学出版社,2009
[124] Yeo.M. V.M, X. P. Li, K Shen, et al. A self-training semi-supervised SVM algorithm andits application in an EEG-based brain computer interface speller system [J]. SafetyScience,2009,47(1):115-124
[125] Jiao. L. C, Liu. Liu. F, J, et al. Intelligent Data Mining and Knowledge Discovery [M]. XiAn: Xidian University press,200
[126] Platt J C, Cristianini N, Shawe-Taylor J. Large margin DAGs for multiclass classification[J]. Advances in neural information processing systems,2000,12(3):547-553
[127] Hsu CW, Lin CJ. A comparison of methods for multi-class support vector machines [J]IEEE Trans. Neural Networks,2002,13(2):415-425
