离线中文签名验证技术研究
|
摘要
签名作为身份确认的一种手段被用于诸多领域,如各类文书、合同、银行票据等。至今为止,围绕自动离线签名验证所展开的研究主要是针对随机和简单模仿两种伪签名展开,尤其是对熟练模仿伪签名的鉴别,还有很多基础理论问题及实际应用问题需要解决。因此开展离线中文签名验证技术研究,具有十分重要的理论意义和实用价值。
围绕模式识别的三个基本研究内容:数据获取和预处理、特征提取和选择以及匹配方法,对离线中文签名验证涉及的若干关键技术进行了概括性讨论。结合噪声消除基本理论以及需关注的主要问题,同时根据实际应用中签名图像中横线像素分布的特点,给出了一种签名图像中横线的清除方法。与常见的噪声消除方法相比,所给出的方法不仅能将对签名图像的影响降低到最低程度,而且易于实现。
计算的复杂度和分类的正确率一直是特征提取关注的两个主要方面。针对中文签字的分类问题,就如何更好体现签名特征这一问题,论述了使用全局特征组成特征矢量的方法。以此为基础,提出了两种模式聚类方法:基于马氏距离的聚类方法和基于特征标权的核聚类方法。
签名分段是签名图像预处理中的一个难点,分段方法的优劣对验证结果的影响极大。在汲取前人在笔划提取方面的成功经验基础之上,同时结合签名验证的特点给出了一种具有较低的时间、空间复杂度,同时具有高鲁棒性的分段方法。此外,还给出了一个行之有效的分段配对和签名相似度计算方法,并在此基础上,给出了模板匹配与RBFNN(Radial-Based Function Neural Network)相结合的离线中文签名验证方法。
应用DHMM (Discrete Hidden Markov Model)进行分类的一个关键问题,是矢量量化的优劣。有鉴于此,基于签名各分段的六维特征矢量,按照其物理意义的不同分为两组分别进行矢量量化。另外,还采用了多个码字表征一个矢量的方法。针对这两种矢量量化改进方法,采用了经调整后的DHMM训练算法。同基于网格方法提取观测值序列相比较,所给出的基于分段获得观测值的方法能更有效的反映中文签名的特点。提出了一种融合GA (Genetic Algorithms)与Baum-Welch两者优点的DHMM训练方法。经GA优化后的HMM在统计意义上提高了验证正确率。
针对熟练模仿伪签名的验证,必须提取更为细节的特征这一基本需求,同时结合签名边缘的像素分布包含更为丰富的签名特征这一基本事实,给出了一种遍历签名边缘的高效算法。通过定义像素的三种运动方向,并由来时方向确定转向方向,该算法能准确指导像素的遍历,确保了遍历的一次完成,使得遍历的时间复杂度降低到O(n) ,从而高效的完成签名边缘的遍历。在签名边缘遍历的基础上,为更为有效的提取到签名的细节特征,探讨了应用小波分析提取二次特征的方法。基于所给出的两种特征提取方法,在HMM (Hidden Markov Model)和SVM (Support Vector Machine)环境下,对离线中文签名验证进行了比较,结果表明HMM更优。
结合一个具代表性的应用需求,设计了一个三级分类器串联融合系统。所构建的系统解决了真实签名样本收集问题。每一级分类器完成一类伪签名的鉴别,并按照由粗糙到精细的顺序排列三级分类器,有效的降低了验证所需的时间。基于真实数据的大批量的测试表明了构建的原型系统能够较好的完成离线自动签名验证任务。
通过对离线中文签名验证技术的研究,取得了若干具理论价值和实用价值的成果,为进一步开展实用系统的研究奠定了理论和方法基础。
As an identity authentication technology, signature verification is used in many kinds offield such as many types of document, business contract and bank cheque et al. Up to the present,researches about automatic off-line signature verification mainly are verifying random forgeriesand simple forgeries. Especially for simulate forgeries, there are many theories and applicationproblems have to be resolved. So, there are great theoretical significance and practical value tocarry out the research of automatic of?ine Chinese signature verification.
Around three basic research contents of pattern recognition: data extraction and preprocess,feature extraction and selection and matcher, several key technologies involved by automaticoff-line signature verification are discussed summarily. Then, in term of the basic theories ofnoise elimination and the key problem that should been concerned when selecting the method ofnoise elimination, and according to the characteristic of pixels layout in line, a horizontal lineclearing method is presented. Compared to generic noise filters, this method can greatly decreasein?uence on the origin signature image and is easier to achieve.
Calculation complexity and correct rate of classifying are the two main aspects concernedby us when selecting feature extraction method. In connection with the classifying of Chinesesignature, to address better re?ecting the characteristics of signature, how to use global featuresto compose the feature vector is discussed. Based on these, two kinds of clustering methods, oneis based on Mahalanobis distance and C-means method, the other is kernel clustering based onfeature-weighting, are presented.
Signature segmentation is a difficult task in preprocess stage. The quality of segmentationwill greatly affect the verification result. A simple, low computational cost and robust segmen-tation method is proposed by means of having successful experiences of strokes extraction forhandwritten Chinese character and taking into account the characteristics of signature verifica-tion. The algorithm presented has a low time and space complexity. Then, a segments sorting andsimilarity degree computation method are presented on the basis of Euclidean distance. Based onthese, a classifier composed with template matching and RBFNN (Radial-Based Function NeuralNetwork) is described.
The quality of vector quantization is a crucial factor when using DHMM (Discrete Hid-den Markov Model) for pattern recognition. In view of this, the feature vectors extracted fromsegments are grouped into two groups according to their physical meaning and each group isquantized respectively. In the vector quantization method, using multi-codeword to represent avector is used. In connection with these two optimization methods of vector quantization, thelearning algorithm of HMM (Hidden Markov Model) is adjusted slightly. Compared to grid method, the method based on stroke segmentation presented in this thesis reserves the layout ofsignature strokes and can extract a better observation sequence. A learning method of DHMM’sparameters which has both excellence of GA (Genetic Algorithm) and Baum-Welch is proposed.The optimized DHMM improves the correct rate in the sense of statistics.
To address the verification of simulated signature, the extraction method of more detailedfeatures is discussed. An efficient tracing algorithm for the big close-ring of signature is pro-posed. Compared to the searching method of tracing route, the tracing algorithm proposed de-creases the time complexity and space complexity greatly. According to the layout of pixels andthe pixel’s direction come from, this tracing algorithm chooses the next pixel, which ensures thetrace process only needs one time and the tracing order consist with the certain order. On thebasis of edge tracing, the extracting method of secondary features is discussed by using waveletpacket transform. The HMM classifier and SVM (support vector machine) classifier are com-pared by using the features extracted. The comparison result shows HMM is better than SVM inour experiments.
At last a cascading combination system composed with three classifiers is built by combin-ing an appropriate application. The verification system solves the problem of genuine signaturecollection. Each classifier verifies one type of forgery. These classifiers are sorted from rough-ness to fine, which decreases the verification time effectively. The promising results of test usinggreat scale actual data indicate the verification system presented in this thesis can finishes theautomatic off-line signature verification task very good.
Through the study on automatic of?ine Chinese signature verification, we achieve sometheory and applied results, and establish the theoretical and methodological basis for the appliedsystem’s research.
围绕模式识别的三个基本研究内容:数据获取和预处理、特征提取和选择以及匹配方法,对离线中文签名验证涉及的若干关键技术进行了概括性讨论。结合噪声消除基本理论以及需关注的主要问题,同时根据实际应用中签名图像中横线像素分布的特点,给出了一种签名图像中横线的清除方法。与常见的噪声消除方法相比,所给出的方法不仅能将对签名图像的影响降低到最低程度,而且易于实现。
计算的复杂度和分类的正确率一直是特征提取关注的两个主要方面。针对中文签字的分类问题,就如何更好体现签名特征这一问题,论述了使用全局特征组成特征矢量的方法。以此为基础,提出了两种模式聚类方法:基于马氏距离的聚类方法和基于特征标权的核聚类方法。
签名分段是签名图像预处理中的一个难点,分段方法的优劣对验证结果的影响极大。在汲取前人在笔划提取方面的成功经验基础之上,同时结合签名验证的特点给出了一种具有较低的时间、空间复杂度,同时具有高鲁棒性的分段方法。此外,还给出了一个行之有效的分段配对和签名相似度计算方法,并在此基础上,给出了模板匹配与RBFNN(Radial-Based Function Neural Network)相结合的离线中文签名验证方法。
应用DHMM (Discrete Hidden Markov Model)进行分类的一个关键问题,是矢量量化的优劣。有鉴于此,基于签名各分段的六维特征矢量,按照其物理意义的不同分为两组分别进行矢量量化。另外,还采用了多个码字表征一个矢量的方法。针对这两种矢量量化改进方法,采用了经调整后的DHMM训练算法。同基于网格方法提取观测值序列相比较,所给出的基于分段获得观测值的方法能更有效的反映中文签名的特点。提出了一种融合GA (Genetic Algorithms)与Baum-Welch两者优点的DHMM训练方法。经GA优化后的HMM在统计意义上提高了验证正确率。
针对熟练模仿伪签名的验证,必须提取更为细节的特征这一基本需求,同时结合签名边缘的像素分布包含更为丰富的签名特征这一基本事实,给出了一种遍历签名边缘的高效算法。通过定义像素的三种运动方向,并由来时方向确定转向方向,该算法能准确指导像素的遍历,确保了遍历的一次完成,使得遍历的时间复杂度降低到O(n) ,从而高效的完成签名边缘的遍历。在签名边缘遍历的基础上,为更为有效的提取到签名的细节特征,探讨了应用小波分析提取二次特征的方法。基于所给出的两种特征提取方法,在HMM (Hidden Markov Model)和SVM (Support Vector Machine)环境下,对离线中文签名验证进行了比较,结果表明HMM更优。
结合一个具代表性的应用需求,设计了一个三级分类器串联融合系统。所构建的系统解决了真实签名样本收集问题。每一级分类器完成一类伪签名的鉴别,并按照由粗糙到精细的顺序排列三级分类器,有效的降低了验证所需的时间。基于真实数据的大批量的测试表明了构建的原型系统能够较好的完成离线自动签名验证任务。
通过对离线中文签名验证技术的研究,取得了若干具理论价值和实用价值的成果,为进一步开展实用系统的研究奠定了理论和方法基础。
As an identity authentication technology, signature verification is used in many kinds offield such as many types of document, business contract and bank cheque et al. Up to the present,researches about automatic off-line signature verification mainly are verifying random forgeriesand simple forgeries. Especially for simulate forgeries, there are many theories and applicationproblems have to be resolved. So, there are great theoretical significance and practical value tocarry out the research of automatic of?ine Chinese signature verification.
Around three basic research contents of pattern recognition: data extraction and preprocess,feature extraction and selection and matcher, several key technologies involved by automaticoff-line signature verification are discussed summarily. Then, in term of the basic theories ofnoise elimination and the key problem that should been concerned when selecting the method ofnoise elimination, and according to the characteristic of pixels layout in line, a horizontal lineclearing method is presented. Compared to generic noise filters, this method can greatly decreasein?uence on the origin signature image and is easier to achieve.
Calculation complexity and correct rate of classifying are the two main aspects concernedby us when selecting feature extraction method. In connection with the classifying of Chinesesignature, to address better re?ecting the characteristics of signature, how to use global featuresto compose the feature vector is discussed. Based on these, two kinds of clustering methods, oneis based on Mahalanobis distance and C-means method, the other is kernel clustering based onfeature-weighting, are presented.
Signature segmentation is a difficult task in preprocess stage. The quality of segmentationwill greatly affect the verification result. A simple, low computational cost and robust segmen-tation method is proposed by means of having successful experiences of strokes extraction forhandwritten Chinese character and taking into account the characteristics of signature verifica-tion. The algorithm presented has a low time and space complexity. Then, a segments sorting andsimilarity degree computation method are presented on the basis of Euclidean distance. Based onthese, a classifier composed with template matching and RBFNN (Radial-Based Function NeuralNetwork) is described.
The quality of vector quantization is a crucial factor when using DHMM (Discrete Hid-den Markov Model) for pattern recognition. In view of this, the feature vectors extracted fromsegments are grouped into two groups according to their physical meaning and each group isquantized respectively. In the vector quantization method, using multi-codeword to represent avector is used. In connection with these two optimization methods of vector quantization, thelearning algorithm of HMM (Hidden Markov Model) is adjusted slightly. Compared to grid method, the method based on stroke segmentation presented in this thesis reserves the layout ofsignature strokes and can extract a better observation sequence. A learning method of DHMM’sparameters which has both excellence of GA (Genetic Algorithm) and Baum-Welch is proposed.The optimized DHMM improves the correct rate in the sense of statistics.
To address the verification of simulated signature, the extraction method of more detailedfeatures is discussed. An efficient tracing algorithm for the big close-ring of signature is pro-posed. Compared to the searching method of tracing route, the tracing algorithm proposed de-creases the time complexity and space complexity greatly. According to the layout of pixels andthe pixel’s direction come from, this tracing algorithm chooses the next pixel, which ensures thetrace process only needs one time and the tracing order consist with the certain order. On thebasis of edge tracing, the extracting method of secondary features is discussed by using waveletpacket transform. The HMM classifier and SVM (support vector machine) classifier are com-pared by using the features extracted. The comparison result shows HMM is better than SVM inour experiments.
At last a cascading combination system composed with three classifiers is built by combin-ing an appropriate application. The verification system solves the problem of genuine signaturecollection. Each classifier verifies one type of forgery. These classifiers are sorted from rough-ness to fine, which decreases the verification time effectively. The promising results of test usinggreat scale actual data indicate the verification system presented in this thesis can finishes theautomatic off-line signature verification task very good.
Through the study on automatic of?ine Chinese signature verification, we achieve sometheory and applied results, and establish the theoretical and methodological basis for the appliedsystem’s research.
引文
[1] Miller B. Vital signs of identity. IEEE Spectrum, 1994, 31(2):22～30
[2] Jain A K, Bolle R, Pankanti S. Biometrics-Personal Identification in Networked Society. 1st ed. Boston:Kluwer Academic Publishers, 1999. 10～15
[3] Pankanti S, Bolle R, Jain A K. Biometrics: The future of identification. Computer, 2000, 33(2):46～80
[4] Jain A K, Ross A, Prabhakar S. An Introduction to Biometric Recognition. IEEE Tans. on Circuits andSystems for Video Technology, 2004, 14(1):4～19
[5] Maltoni D, Maio D, Jain A K, et al. Handbook of Fingerprint Recognition. New York: Springer-Verlag,2003
[6] 孙冬梅, 裘正定. 生物特征识别技术综述. 电子学报, 2001, 29(12A):1744～1748
[7] Kong S, Heo J, Abidi B, et al. Recent Advances in Visual and Infrared Face Recognition-A Review.Computer Vision and Image Understanding, 2005, 97(1):103～135
[8] Kumar A, Wong D C, Shen H C, et al. Personal verification using palmprint and hand geometry biometric.in: Kasturi R, Laurendeau D, editors. Proceedings of the 4th International Conference on Audio-andVideo-based Biometric Person Authentication. Los, 2003. 320～327
[9] Zhang D, Shu W. Two novel characteristic in palmprint verification: Datum point invariance and linefeature matching. Pattern Recognition, 1999, 32(4):691～702
[10] Daugman J. How Iris Recognition Works. IEEE Trans. on Circuits and Systems for Video Technology,2004, 14(1):21～30
[11] Rosenberg A E. Automatic Speaker Verification: A Review. Proceedings of the IEEE, 1976,64(4):475～487
[12] Huang K, Yan H. Off-line signature verification using structural feature correspondence. Pattern Recog-nition, 2002, 35(11):2467～2477
[13] Hanmandlua M, Mohd.Yusofb M H, Madasuc V K. Off-line signature verification and forgery detectionusing fuzzy modeling. Pattern Recognition, 2005, 38(3):341～356
[14] Yoshiki M, Mitsu Y, Hidetoshi M, et al. An off-line signature verification system using an extracteddisplacement function. Pattern Recognition Letter, 2002, 23(13):1569～1577
[15] Leclerc F, Plamondon R. Understanding Users’ Keystroke Patterns for Computer Access Security. Com-puters & Security, 2003, 22(8):695～706
[16] Kresimir D, Mislav G. A Survery of Biometric Recognition Methods. in: GrgiE T K M, editor. Proceed-ings of the 46th International Symposium Electronics in Marine. Zadar, Croatia: IEEE Computer ScietyPress, 2004. 184～193
[17] 孙哲南, 谭铁牛. 生物特征识别技术的研究现状和发展趋势. 见: 中国计算机科学技术发展报告2005. 北京: 清华大学出版社, 2006. 215～235
[18] Daigo Muramatsu M S S T T M. A Markov Chain Monte Carlo Algorithm for Bayesian DynamicSignature Verification. IEEE Transactions on Information Forensics and Security, 2006, 1(1):22～34
[19] Plamondon R, Lorette G. Automatic signature verification and writer identification: The state of the art.Pattern Recognition, 1989, 22(2):107～131
[20] Leclerc F, Plamondon R. Automatic signature verification: The state of the art 1989-1993. InternationalJournal of Pattern Recognition and Artificial Intelligence, 1994, 8(3):643～660
[21] Dimauro G, Impedovo S, Lucchese M G, et al. Recent advancements in automatic signature verification.in: Proceedings of 9th International Workshop on Frontiers in Handwriting Recognition. Tokyo, Japan:IEEE Computer Sciety Press, 2004. 179～184
[22] Weiping Hou K W. A Survey of Off-line Signature Verification. in: Proceedings of the 2004 InternationalConference on Intelligent Mechatronics and Automation. Chengdu, China: IEEE Computer Sciety, 2004.536～541
[23] Gonzalez R C, Woods R E. Digital Image Processing. 2nd ed. Upper Saddle River: Prentice Hall, 2002.35～107
[24] Gonzalez R C, Woods R E, Eddins S L. Digital Image Processing Using Matlab. 1st ed. Upper SaddleRiver: Prentice Hall, 2004. 353～355
[25] G.Dimauro, Impedovo, Pirlo G. Off-line Signature Verification through Fundamental Strokes Analysis.in: S.Impedovo, editor. Proceedings of in Progress in Image Analysis and Processing III. Singapore andLondon: World Scientific, 1994. 331～337
[26] Sabourin R, Plamondon R. Segmentation of Handwritten Signature Images Using the Statistics of Direc-tional Data. in: Proceedings of 9th ICPR. Rome, Italy: IEEE Computer Society Press, 1988. 282～285
[27] Ammar M, Yoshida Y, Fukumura T. Structural Description Classification of Signature Images. PatternRecognition, 1990, 23(7):697～710
[28] Sabourin R, Drouhard J. Off-line signature verification using directional PDF and neural networks. in:Backer E, Gelsema E, editors. Proceedings of 11th ICPR. The Hague, The Netherlands: IEEE ComputerSociety Press, 1992. 321～325
[29] Baltzakis H, Papamarkos N. A new signature verification technique based on a two-stage neural networkclassifier. Engineering Application of AI, 2001, 14(1):95～103
[30] Sabourin R, Plamondon R, Beaumier L. Structural interpretation of handwritten signature images. Inter-national Journal of Pattern Recognition and Artificial Intelligence, 1994, 8(3):709～748
[31] Bajaj R, Chaudhury S. Signature Verification using multiple neural classifiers. Pattern Recognition,1997, 30(1):1～7
[32] Cardot H, Revenu M, Victorri B, et al. A Static Signature Verification System based on a cooperativeNeural Networks Architecture. International Journal of Pattern Recognition and Artificial Intelligence,1994, 8(3):679～692
[33] Hai L, Hai Z L. Chinese signature verification with moment invariants. in: Proceedings of IEEE Inter-national Conference on System, Man and Cybernetics. Piscataway, NJ, USA: IEEE, 1996. 2963～2966
[34] 左文明, 黎绍发, 曾宪贵. 基于小波变换和不变矩的脱机签名鉴别. 微电子学与计算机, 2004,21(12):24～27
[35] Deng P S, Liao H Y M, Ho C W, et al. Wavelet-based Off-line Signature Verification. Computer Visionand Image Understanding, 1999, 76(3):173～190
[36] Nemcek W F, Lin W C. Experimental investigation of Model automatic signature verification. IEEET-SMC, 1974, 4(1):121～126
[37] Yoshimara I, Yoshimura M, Tsukamoto T. Investigation of an Automatic Verification System for JapaneseCountersignatuers on Traver’s Cheques. in: Simner M, Hulstijn W, Girouard P, editors. Proceedings of7th Biennial Conference Of The International Graphonomics Society. Amsterdam, 1995. 86～87
[38] Ueda K. Investigation of Off-line Japanese Signature Verification Using a Pattern Matching. in: Pro-ceedings of the Seventh International Conference on Document Analysis and Recognition. Edinburgh,Scotland, UK: IEEE Computer Sciety Press, 2003. 951～955
[39] Jain A K, Duin R P W, Mao J. Statistical Pattern Recognition: A Review. IEEE Transactions on PatternAnalysis and Machine Intelligence, 2000, 22(1):4～37
[40] Fang B, Leung C H, Tang Y Y, et al. Off-line signature verication by the tracking of feature and strokepositions. Pattern Recognition, 2003, 36(1):91～101
[41] Guo J K, Doermann D, Rosenfeld A. Forgery Detection by Local Correspondence. International Journalof Pattern Recognition and Artificial Intelligence, 2001, 15(1):579～641
[42] McKeague I W. A statistical model for signature verification. Journal of the American Statistical Asso-ciation, 2005, 100(469):231～241
[43] Justino E, Bortolozzi F, Sabourin R. Off-line Signature Verification Using HMM for Random, Simpleand Skilled Forgeries. in: Proceedings of the ICDAR 2001. Seattle, WA, USA: IEEE Computer ScietyPress, 2001. 105～110
[44] Bishop C M. Neural Networks for Pattern Recognition. New York, NY, USA: Oxford University Press,1995
[45] Mighell D A, Wilkinson T S, Goodman J W. Backpropagation and Its Application to Handwritten Sig-nature Verification. in: Proceedings of Advances in Neural Information Processing System. CambridgeMA: the MIT Press, 1989. 340～347
[46] McCormack D K, Brown B M, Pedersen J F. Neural network signature verification using Haar waveletand Fourier transforms. in: Batchelor B G, Solomon S S, Waltz F M, editors. Proceedings of MachineVision Applications, Architectures, and Systems Integration II. Boston MA: SPIE, 1993. 14～25
[47] Drouhard J P, Sabourin R, Godbout M. A Neural Network Approach to Off-line Signature VerificationUsing Directional PDF. Pattern Recognition, 1996, 29(3):415～424
[48] Huang K, Yan H. Off-line Signature Verification Based on Geometric Feature Extraction and NeuralNetwork Classification. Pattern Recognition, 1997, 30(1):9～17
[49] Abbas R. Backpropagation Networks Prototype for Of?ine Signature Verification: [Master Thesis].Cambridge, MA: Massachusetts Institute of Technology, 1994
[50] Chuang P C. Machine Verification of Handwritten Signature Image. in: Jackson J, Vore R W D, editors.Proceedings of International Conference On Crime Countermeasures-Sci. University of Kentucky, 1977.105～109
[51] Ammar M, Yoshida Y, Fukumura T. A New Effective Approach for Off-line Verification of Signaturesby Using Pressure Features. in: Proceedings of The 8th International Conference on Pattern Recognition.Paris, France, 1986. 176～184
[52] Ammar M. Process in verification of skillfully simulated handwritten signature images. InternationalJournal of Pattern Recognition and Artificial Intelligence, 1991, 2(4):589～602
[53] Lam L, Lee S W, Suen C Y. Thinning Methodologies-A Comprehensive Survey. IEEE Transactions onPattern Analysis and Machine Intelligence, 1992, 9(14):869～885
[54] Belkasim O, Shridhar M, Ahmadi M. Pattern recognition with moment invariants: a comparative studyand new result. Patten Recognition, 1991, 24(12):1117～1138
[55] Beiley R R, Srinath M. Orthogonal moment features for use with parametric and non-parametric classi-fiers. IEEE Transaction on Patten Analysis and Machine Intelligent, 1998, 18(4):389～399
[56] Mallat S G. Zero crossings of a wavelet transform. IEEE Transactions on Information Theory, 1991,37(4):1019～1033
[57] 左文明, 黎绍发, 曾宪贵. 脱机手写签名纹理特征分析. 微电子学与计算机, 2005, 22(2):154～157
[58] 林峻, 李介谷. 离线中文签名鉴别的特征提取及预处理. 上海交通大学学报, 1996, 30(9):112～118
[59] Jain A, Zongker D. Feature Selection: Evaluation, Application, and Small Sample Performance. IEEETransactions on Pattern Analysis and Machine Intelligence, 1997, 19(2):153～158
[60] Dash M, Liu H. Feature Selection for Classification. Intelligent Data Analysis, 1997, 1(3):131～156
[61] Yang Y, Pedersen J O. A comparative study on feature selection in text categorization. in: Fisher D H,editor. Proceedings of ICML-97, 14th International Conference on Machine Learning. San Francisco,US: Morgan Kaufmann, 1997. 412～420
[62] Cunha Cavalcanti G D, Doria R, Carvalho E. Feature Selection For Off-Line Recognition of Differ-ent Size Signatures. in: Proceedings of IEEE International Workshop on Neural Networks for SignalProcessing (NNSP 2002). Martigny,Valais-Switzerland: IEEE Computer Sciety Press, 2002. 355～364
[63] Vapnik V N. The Nature of Statistical Learning. 2nd ed. Berlin: Springer, 1995. 94～97
[64] Vapnik V N. Statistical learning theory. 1st ed. New York: John Wiley And Sons, 1998. 103～105
[65] BURGES C J. A tutorial on Support Vector Machines for pattern recognition. Knowledge Discovery andData Mining, 1998, 2(2):121～167
[66] Cristianini N, Shawe-Taylor J. An Introduction to Support Vector Machine and Other Kernel-basedLearning Methods. 1st ed. Cambridge: Cambridge University Press, 2000. 70～80
[67] Hu M. Visual pattern recognition by moment invariants. IEEE Transactions on Information Theory,1962, 8(2):179～187
[68] 张莉, 周伟达, 焦李成. 核聚类算法. 计算机学报, 2002, 25(6):587～590
[69] Hussein F, Kharma N, Ward R. Genetic algorithms for feature selection and weighting, a review andstudy. in: Proceedings of International Conference on Document Analysis and Recognition. Seattle,WA, USA: IEEE Computer Sciety Press, 2001. 176～184
[70] Jayanta B, Rajat K D, Sankar K P. Unsupervised feature selection using a neuro-fuzzy approach. PatternRecognition Letters, 1998, 19(11):997～1006
[71] 贝贵琴, 张学涛. 汉字频度统计. 北京: 电子工业出版社, 1988
[72] Li J, Gao X, Ji H. A feature weighted FCM clustering algorithm based on evolutionary strategy. in: JHC, P L, Y C, editors. Proceedings of the 4th world congress on intelligent and automation. Shanghai,China: Morgan Kaufmann Publishers, 2002. 1549～1553
[73] Abuhaiba I S I, Holt M J J, Datta S. Processing of Binary Images of Handwritten Text Documents.Pattern Recognition, 1996, 29(7):1161～1177
[74] Liu K, Huang Y S, Suen C Y. Robust Stroke Segmentation Method for Handwritten Chinese CharacterRecognition. in: Kasturi R, Laurendeau D, editors. Proceedings of the Fourth International Conferenceon Document Analysis and Recognition. Los Alamitos, USA: IEEE Computer Society, 1997. 211～215
[75] Lin F, Tang X. Of?ine Handwritten Chinese Character Stroke Extraction. in: Kasturi R, Laurendeau D,Suen C, editors. Proceedings of 16th International Conference on Pattern Recognition. Los Alamitos,USA: IEEE Computer Society, 2002. 249～252
[76] 向日华, 王润生. 一种基于高斯混合模型的距离图像分割算法. 软件学报, 2003, 14(7):1250～1257
[77] Rabiner L R. A Tutorial on Hidden Markov Models and Selected Applications in Speech Recognition.Proceedings of the IEEE, 1989, 77(2):257～286
[78] Senior A W, Robinson A J. An Off-line Cursive Handwriting Recognition System. IEEE Transactionson Pattern Analysis and Machine Intelligence, 1998, 20(3):309～321
[79] Liu X, Wan D, Zhang L. An Approach for Face Recognition Based on Singular Value Decompositionand Hidden Markov Model. Chinese Journal of Computers, 2003, 26(3):340～344
[80] Kashi R S, Hu J, Nelson W L, et al. On-line Handwritten Signature Verification Using Hidden MarkovModel Features. in: Kasturi R, Laurendeau D, editors. Proceedings of the Fourth International Con-ference on Document Analysis and Recognition. Los Alamitos, USA: IEEE Computer Society, 1997.253～257
[81] Martens R, Claesen L. Incorporating Local Consistency Information into the On-line Signature Verifica-tion Process. International Journal on Document Analysis and Recognition, 1998, 1(2):110～115
[82] Kashi R, Hu J, Nelson W, et al. A Hidden Markov Model Approach to Online Handwritten SignatureVerification. International Journal on Document Analysis and Recognition, 1998, 1(2):102～109
[83] Yang L, Widjaja B K, Prasad R. Application of Hidden Markov Models for Signature Verification. PatternRecognition, 1995, 28(2):161～170
[84] BILMES J A. What HMMs Can Do. IEICE Transactions on Information and Systems, 2006, E89-D(3):869～891
[85] Justino E, Bortolozzi F, Sabourin R. An Off-line Signature Verification System Using Hidden MarkovModel and Cross-Validation. in: Gramado B, editor. Proceedings of XIII Brazilian Symposium on Com-puter Graphics and Image Processing. Los Alamitos, USA: IEEE Computer Society, 2000. 105～112
[86] Baum L E, Petrie T, Soules G, et al. A maximization technique occurring in the statistical analysis ofprobabilistic functions of Markov chains. Ann. Math. Stat., 1970, 41(1):164～171
[87] Baum L E. An inequality and associated maximization technique in statistical estimation for probabilisticfunctions of Markov processes. Inequalities, 1972, 3(1):1～8
[88] Baker J K. The Dragon System-An Overview. IEEE Trans. Acoust Speech Signal Processing, 1975,ASSP-23(1):24～29
[89] Jelinek F, Bahl L R, Mercer R L. Design of a Linguistic Statistical Decoder for the Recognition ofContinuous Speech. IEEE Trans. Information Theory, 1975, IT-21(1):404～411
[90] Rabiner L R, Juang B H. An Introduction to Hidden Markov Models. IEEE Trans. Acoust Speech SignalProcessing Mag., 1986, 3(1):4～16
[91] Chau C W, Kwong S, Diu C K, et al. Optimization of HMM by a Genetic Algorithm. in: Proceedings ofthe IEEE International Conference on acoustics, Speech, and Signal Processing. Philadelphia, PA, USA:IEEE Computer Society, 1997. 1727～1730
[92] Won K J, Prügel-Bennett A, Krogh A. Evolving the Structure of hidden Markov Models. IEEE Transac-tions on Evolutionary Computation, 2006, 10(1):19～29
[93] Michalewicz Z. Genetic Algorithms + Data Structures = Evolution Programs. 3nd ed. New York:Springer-Verlag, 1996. 96～102
[94] Soille P. Morphological Image Analysis: Principles and Applications. 2nd ed. Berlin: Springer-Verlag,1998. 63～135
[95] 刘毅, 张彩明, 赵玉华, 等. 基于多尺度小波包分析的肺音特征提取与分类. 计算机学报, 2006,29(5):769～777
[96] Xu L, Krzyzak A, Suen C. Methods for Combining Multiple Classifiers and Their Applications inHandwritten Character Recognition. IEEE Trans. Systems, Man, and Cybernetics, 1992, 22(3):418～435
[2] Jain A K, Bolle R, Pankanti S. Biometrics-Personal Identification in Networked Society. 1st ed. Boston:Kluwer Academic Publishers, 1999. 10～15
[3] Pankanti S, Bolle R, Jain A K. Biometrics: The future of identification. Computer, 2000, 33(2):46～80
[4] Jain A K, Ross A, Prabhakar S. An Introduction to Biometric Recognition. IEEE Tans. on Circuits andSystems for Video Technology, 2004, 14(1):4～19
[5] Maltoni D, Maio D, Jain A K, et al. Handbook of Fingerprint Recognition. New York: Springer-Verlag,2003
[6] 孙冬梅, 裘正定. 生物特征识别技术综述. 电子学报, 2001, 29(12A):1744～1748
[7] Kong S, Heo J, Abidi B, et al. Recent Advances in Visual and Infrared Face Recognition-A Review.Computer Vision and Image Understanding, 2005, 97(1):103～135
[8] Kumar A, Wong D C, Shen H C, et al. Personal verification using palmprint and hand geometry biometric.in: Kasturi R, Laurendeau D, editors. Proceedings of the 4th International Conference on Audio-andVideo-based Biometric Person Authentication. Los, 2003. 320～327
[9] Zhang D, Shu W. Two novel characteristic in palmprint verification: Datum point invariance and linefeature matching. Pattern Recognition, 1999, 32(4):691～702
[10] Daugman J. How Iris Recognition Works. IEEE Trans. on Circuits and Systems for Video Technology,2004, 14(1):21～30
[11] Rosenberg A E. Automatic Speaker Verification: A Review. Proceedings of the IEEE, 1976,64(4):475～487
[12] Huang K, Yan H. Off-line signature verification using structural feature correspondence. Pattern Recog-nition, 2002, 35(11):2467～2477
[13] Hanmandlua M, Mohd.Yusofb M H, Madasuc V K. Off-line signature verification and forgery detectionusing fuzzy modeling. Pattern Recognition, 2005, 38(3):341～356
[14] Yoshiki M, Mitsu Y, Hidetoshi M, et al. An off-line signature verification system using an extracteddisplacement function. Pattern Recognition Letter, 2002, 23(13):1569～1577
[15] Leclerc F, Plamondon R. Understanding Users’ Keystroke Patterns for Computer Access Security. Com-puters & Security, 2003, 22(8):695～706
[16] Kresimir D, Mislav G. A Survery of Biometric Recognition Methods. in: GrgiE T K M, editor. Proceed-ings of the 46th International Symposium Electronics in Marine. Zadar, Croatia: IEEE Computer ScietyPress, 2004. 184～193
[17] 孙哲南, 谭铁牛. 生物特征识别技术的研究现状和发展趋势. 见: 中国计算机科学技术发展报告2005. 北京: 清华大学出版社, 2006. 215～235
[18] Daigo Muramatsu M S S T T M. A Markov Chain Monte Carlo Algorithm for Bayesian DynamicSignature Verification. IEEE Transactions on Information Forensics and Security, 2006, 1(1):22～34
[19] Plamondon R, Lorette G. Automatic signature verification and writer identification: The state of the art.Pattern Recognition, 1989, 22(2):107～131
[20] Leclerc F, Plamondon R. Automatic signature verification: The state of the art 1989-1993. InternationalJournal of Pattern Recognition and Artificial Intelligence, 1994, 8(3):643～660
[21] Dimauro G, Impedovo S, Lucchese M G, et al. Recent advancements in automatic signature verification.in: Proceedings of 9th International Workshop on Frontiers in Handwriting Recognition. Tokyo, Japan:IEEE Computer Sciety Press, 2004. 179～184
[22] Weiping Hou K W. A Survey of Off-line Signature Verification. in: Proceedings of the 2004 InternationalConference on Intelligent Mechatronics and Automation. Chengdu, China: IEEE Computer Sciety, 2004.536～541
[23] Gonzalez R C, Woods R E. Digital Image Processing. 2nd ed. Upper Saddle River: Prentice Hall, 2002.35～107
[24] Gonzalez R C, Woods R E, Eddins S L. Digital Image Processing Using Matlab. 1st ed. Upper SaddleRiver: Prentice Hall, 2004. 353～355
[25] G.Dimauro, Impedovo, Pirlo G. Off-line Signature Verification through Fundamental Strokes Analysis.in: S.Impedovo, editor. Proceedings of in Progress in Image Analysis and Processing III. Singapore andLondon: World Scientific, 1994. 331～337
[26] Sabourin R, Plamondon R. Segmentation of Handwritten Signature Images Using the Statistics of Direc-tional Data. in: Proceedings of 9th ICPR. Rome, Italy: IEEE Computer Society Press, 1988. 282～285
[27] Ammar M, Yoshida Y, Fukumura T. Structural Description Classification of Signature Images. PatternRecognition, 1990, 23(7):697～710
[28] Sabourin R, Drouhard J. Off-line signature verification using directional PDF and neural networks. in:Backer E, Gelsema E, editors. Proceedings of 11th ICPR. The Hague, The Netherlands: IEEE ComputerSociety Press, 1992. 321～325
[29] Baltzakis H, Papamarkos N. A new signature verification technique based on a two-stage neural networkclassifier. Engineering Application of AI, 2001, 14(1):95～103
[30] Sabourin R, Plamondon R, Beaumier L. Structural interpretation of handwritten signature images. Inter-national Journal of Pattern Recognition and Artificial Intelligence, 1994, 8(3):709～748
[31] Bajaj R, Chaudhury S. Signature Verification using multiple neural classifiers. Pattern Recognition,1997, 30(1):1～7
[32] Cardot H, Revenu M, Victorri B, et al. A Static Signature Verification System based on a cooperativeNeural Networks Architecture. International Journal of Pattern Recognition and Artificial Intelligence,1994, 8(3):679～692
[33] Hai L, Hai Z L. Chinese signature verification with moment invariants. in: Proceedings of IEEE Inter-national Conference on System, Man and Cybernetics. Piscataway, NJ, USA: IEEE, 1996. 2963～2966
[34] 左文明, 黎绍发, 曾宪贵. 基于小波变换和不变矩的脱机签名鉴别. 微电子学与计算机, 2004,21(12):24～27
[35] Deng P S, Liao H Y M, Ho C W, et al. Wavelet-based Off-line Signature Verification. Computer Visionand Image Understanding, 1999, 76(3):173～190
[36] Nemcek W F, Lin W C. Experimental investigation of Model automatic signature verification. IEEET-SMC, 1974, 4(1):121～126
[37] Yoshimara I, Yoshimura M, Tsukamoto T. Investigation of an Automatic Verification System for JapaneseCountersignatuers on Traver’s Cheques. in: Simner M, Hulstijn W, Girouard P, editors. Proceedings of7th Biennial Conference Of The International Graphonomics Society. Amsterdam, 1995. 86～87
[38] Ueda K. Investigation of Off-line Japanese Signature Verification Using a Pattern Matching. in: Pro-ceedings of the Seventh International Conference on Document Analysis and Recognition. Edinburgh,Scotland, UK: IEEE Computer Sciety Press, 2003. 951～955
[39] Jain A K, Duin R P W, Mao J. Statistical Pattern Recognition: A Review. IEEE Transactions on PatternAnalysis and Machine Intelligence, 2000, 22(1):4～37
[40] Fang B, Leung C H, Tang Y Y, et al. Off-line signature verication by the tracking of feature and strokepositions. Pattern Recognition, 2003, 36(1):91～101
[41] Guo J K, Doermann D, Rosenfeld A. Forgery Detection by Local Correspondence. International Journalof Pattern Recognition and Artificial Intelligence, 2001, 15(1):579～641
[42] McKeague I W. A statistical model for signature verification. Journal of the American Statistical Asso-ciation, 2005, 100(469):231～241
[43] Justino E, Bortolozzi F, Sabourin R. Off-line Signature Verification Using HMM for Random, Simpleand Skilled Forgeries. in: Proceedings of the ICDAR 2001. Seattle, WA, USA: IEEE Computer ScietyPress, 2001. 105～110
[44] Bishop C M. Neural Networks for Pattern Recognition. New York, NY, USA: Oxford University Press,1995
[45] Mighell D A, Wilkinson T S, Goodman J W. Backpropagation and Its Application to Handwritten Sig-nature Verification. in: Proceedings of Advances in Neural Information Processing System. CambridgeMA: the MIT Press, 1989. 340～347
[46] McCormack D K, Brown B M, Pedersen J F. Neural network signature verification using Haar waveletand Fourier transforms. in: Batchelor B G, Solomon S S, Waltz F M, editors. Proceedings of MachineVision Applications, Architectures, and Systems Integration II. Boston MA: SPIE, 1993. 14～25
[47] Drouhard J P, Sabourin R, Godbout M. A Neural Network Approach to Off-line Signature VerificationUsing Directional PDF. Pattern Recognition, 1996, 29(3):415～424
[48] Huang K, Yan H. Off-line Signature Verification Based on Geometric Feature Extraction and NeuralNetwork Classification. Pattern Recognition, 1997, 30(1):9～17
[49] Abbas R. Backpropagation Networks Prototype for Of?ine Signature Verification: [Master Thesis].Cambridge, MA: Massachusetts Institute of Technology, 1994
[50] Chuang P C. Machine Verification of Handwritten Signature Image. in: Jackson J, Vore R W D, editors.Proceedings of International Conference On Crime Countermeasures-Sci. University of Kentucky, 1977.105～109
[51] Ammar M, Yoshida Y, Fukumura T. A New Effective Approach for Off-line Verification of Signaturesby Using Pressure Features. in: Proceedings of The 8th International Conference on Pattern Recognition.Paris, France, 1986. 176～184
[52] Ammar M. Process in verification of skillfully simulated handwritten signature images. InternationalJournal of Pattern Recognition and Artificial Intelligence, 1991, 2(4):589～602
[53] Lam L, Lee S W, Suen C Y. Thinning Methodologies-A Comprehensive Survey. IEEE Transactions onPattern Analysis and Machine Intelligence, 1992, 9(14):869～885
[54] Belkasim O, Shridhar M, Ahmadi M. Pattern recognition with moment invariants: a comparative studyand new result. Patten Recognition, 1991, 24(12):1117～1138
[55] Beiley R R, Srinath M. Orthogonal moment features for use with parametric and non-parametric classi-fiers. IEEE Transaction on Patten Analysis and Machine Intelligent, 1998, 18(4):389～399
[56] Mallat S G. Zero crossings of a wavelet transform. IEEE Transactions on Information Theory, 1991,37(4):1019～1033
[57] 左文明, 黎绍发, 曾宪贵. 脱机手写签名纹理特征分析. 微电子学与计算机, 2005, 22(2):154～157
[58] 林峻, 李介谷. 离线中文签名鉴别的特征提取及预处理. 上海交通大学学报, 1996, 30(9):112～118
[59] Jain A, Zongker D. Feature Selection: Evaluation, Application, and Small Sample Performance. IEEETransactions on Pattern Analysis and Machine Intelligence, 1997, 19(2):153～158
[60] Dash M, Liu H. Feature Selection for Classification. Intelligent Data Analysis, 1997, 1(3):131～156
[61] Yang Y, Pedersen J O. A comparative study on feature selection in text categorization. in: Fisher D H,editor. Proceedings of ICML-97, 14th International Conference on Machine Learning. San Francisco,US: Morgan Kaufmann, 1997. 412～420
[62] Cunha Cavalcanti G D, Doria R, Carvalho E. Feature Selection For Off-Line Recognition of Differ-ent Size Signatures. in: Proceedings of IEEE International Workshop on Neural Networks for SignalProcessing (NNSP 2002). Martigny,Valais-Switzerland: IEEE Computer Sciety Press, 2002. 355～364
[63] Vapnik V N. The Nature of Statistical Learning. 2nd ed. Berlin: Springer, 1995. 94～97
[64] Vapnik V N. Statistical learning theory. 1st ed. New York: John Wiley And Sons, 1998. 103～105
[65] BURGES C J. A tutorial on Support Vector Machines for pattern recognition. Knowledge Discovery andData Mining, 1998, 2(2):121～167
[66] Cristianini N, Shawe-Taylor J. An Introduction to Support Vector Machine and Other Kernel-basedLearning Methods. 1st ed. Cambridge: Cambridge University Press, 2000. 70～80
[67] Hu M. Visual pattern recognition by moment invariants. IEEE Transactions on Information Theory,1962, 8(2):179～187
[68] 张莉, 周伟达, 焦李成. 核聚类算法. 计算机学报, 2002, 25(6):587～590
[69] Hussein F, Kharma N, Ward R. Genetic algorithms for feature selection and weighting, a review andstudy. in: Proceedings of International Conference on Document Analysis and Recognition. Seattle,WA, USA: IEEE Computer Sciety Press, 2001. 176～184
[70] Jayanta B, Rajat K D, Sankar K P. Unsupervised feature selection using a neuro-fuzzy approach. PatternRecognition Letters, 1998, 19(11):997～1006
[71] 贝贵琴, 张学涛. 汉字频度统计. 北京: 电子工业出版社, 1988
[72] Li J, Gao X, Ji H. A feature weighted FCM clustering algorithm based on evolutionary strategy. in: JHC, P L, Y C, editors. Proceedings of the 4th world congress on intelligent and automation. Shanghai,China: Morgan Kaufmann Publishers, 2002. 1549～1553
[73] Abuhaiba I S I, Holt M J J, Datta S. Processing of Binary Images of Handwritten Text Documents.Pattern Recognition, 1996, 29(7):1161～1177
[74] Liu K, Huang Y S, Suen C Y. Robust Stroke Segmentation Method for Handwritten Chinese CharacterRecognition. in: Kasturi R, Laurendeau D, editors. Proceedings of the Fourth International Conferenceon Document Analysis and Recognition. Los Alamitos, USA: IEEE Computer Society, 1997. 211～215
[75] Lin F, Tang X. Of?ine Handwritten Chinese Character Stroke Extraction. in: Kasturi R, Laurendeau D,Suen C, editors. Proceedings of 16th International Conference on Pattern Recognition. Los Alamitos,USA: IEEE Computer Society, 2002. 249～252
[76] 向日华, 王润生. 一种基于高斯混合模型的距离图像分割算法. 软件学报, 2003, 14(7):1250～1257
[77] Rabiner L R. A Tutorial on Hidden Markov Models and Selected Applications in Speech Recognition.Proceedings of the IEEE, 1989, 77(2):257～286
[78] Senior A W, Robinson A J. An Off-line Cursive Handwriting Recognition System. IEEE Transactionson Pattern Analysis and Machine Intelligence, 1998, 20(3):309～321
[79] Liu X, Wan D, Zhang L. An Approach for Face Recognition Based on Singular Value Decompositionand Hidden Markov Model. Chinese Journal of Computers, 2003, 26(3):340～344
[80] Kashi R S, Hu J, Nelson W L, et al. On-line Handwritten Signature Verification Using Hidden MarkovModel Features. in: Kasturi R, Laurendeau D, editors. Proceedings of the Fourth International Con-ference on Document Analysis and Recognition. Los Alamitos, USA: IEEE Computer Society, 1997.253～257
[81] Martens R, Claesen L. Incorporating Local Consistency Information into the On-line Signature Verifica-tion Process. International Journal on Document Analysis and Recognition, 1998, 1(2):110～115
[82] Kashi R, Hu J, Nelson W, et al. A Hidden Markov Model Approach to Online Handwritten SignatureVerification. International Journal on Document Analysis and Recognition, 1998, 1(2):102～109
[83] Yang L, Widjaja B K, Prasad R. Application of Hidden Markov Models for Signature Verification. PatternRecognition, 1995, 28(2):161～170
[84] BILMES J A. What HMMs Can Do. IEICE Transactions on Information and Systems, 2006, E89-D(3):869～891
[85] Justino E, Bortolozzi F, Sabourin R. An Off-line Signature Verification System Using Hidden MarkovModel and Cross-Validation. in: Gramado B, editor. Proceedings of XIII Brazilian Symposium on Com-puter Graphics and Image Processing. Los Alamitos, USA: IEEE Computer Society, 2000. 105～112
[86] Baum L E, Petrie T, Soules G, et al. A maximization technique occurring in the statistical analysis ofprobabilistic functions of Markov chains. Ann. Math. Stat., 1970, 41(1):164～171
[87] Baum L E. An inequality and associated maximization technique in statistical estimation for probabilisticfunctions of Markov processes. Inequalities, 1972, 3(1):1～8
[88] Baker J K. The Dragon System-An Overview. IEEE Trans. Acoust Speech Signal Processing, 1975,ASSP-23(1):24～29
[89] Jelinek F, Bahl L R, Mercer R L. Design of a Linguistic Statistical Decoder for the Recognition ofContinuous Speech. IEEE Trans. Information Theory, 1975, IT-21(1):404～411
[90] Rabiner L R, Juang B H. An Introduction to Hidden Markov Models. IEEE Trans. Acoust Speech SignalProcessing Mag., 1986, 3(1):4～16
[91] Chau C W, Kwong S, Diu C K, et al. Optimization of HMM by a Genetic Algorithm. in: Proceedings ofthe IEEE International Conference on acoustics, Speech, and Signal Processing. Philadelphia, PA, USA:IEEE Computer Society, 1997. 1727～1730
[92] Won K J, Prügel-Bennett A, Krogh A. Evolving the Structure of hidden Markov Models. IEEE Transac-tions on Evolutionary Computation, 2006, 10(1):19～29
[93] Michalewicz Z. Genetic Algorithms + Data Structures = Evolution Programs. 3nd ed. New York:Springer-Verlag, 1996. 96～102
[94] Soille P. Morphological Image Analysis: Principles and Applications. 2nd ed. Berlin: Springer-Verlag,1998. 63～135
[95] 刘毅, 张彩明, 赵玉华, 等. 基于多尺度小波包分析的肺音特征提取与分类. 计算机学报, 2006,29(5):769～777
[96] Xu L, Krzyzak A, Suen C. Methods for Combining Multiple Classifiers and Their Applications inHandwritten Character Recognition. IEEE Trans. Systems, Man, and Cybernetics, 1992, 22(3):418～435