摘要
信息融合技术已经成功地应用于众多的研究领域，在模式识别领域也有巨大的应用价值。通过研究可以发现，对于复杂模式识别问题，如手写体汉字识别，可以说目前还没有一个简单的方法可以达到较高的识别率和可靠度，每一种方法都有各自的优点、缺陷和不同的适用范围，不同的特征和匹配方法之间具有一定的互补性。因此，研究如何将不同的方法有机地结合起来以充分发挥各自的优势，克服其缺陷，从而构成信息融合型的识别系统，就成为当前模式识别研究的一个主要方向。本文主要在基于信息融合结构的手写体汉字识别理论和应用方面进行了以下工作：
1、手写体汉字变形问题是手写体汉字识别中的关键问题，归一化处理是直接从汉字点阵图像上矫正手写变形、减小属于同一类别的不同模式之间差异的途径。通过对目前文献中的几种归一化处理算法的深入研究，提出了一种新的非线性归一化方法。该方法在进行空间坐标变换归一化处理时，采取线密度填充的算法，使得笔划密度的描述更为合理，归一化后的汉字点阵中笔划的分布更加均匀。
2、汉字特征直接反映着汉字形体整体或局部分布状况，良好的特征应该使同一种汉字的不同书写样本之间的差异性尽可能小,而在不同汉字之间的差异性尽可能大。根据手写体汉字的特点，对特征提取进行了分析研究，提出了改进方向线素特征、笔划分区矩特征、扩展周边笔划方向特征等几种新的手写体汉字特征，实验表明所提取的几种特征是有效的。
3、研究了基于自适应特征融合及模块神经网络的手写体汉字识别。在传统的特征提取方法基础上, 给出了多特征融合的一种方法，多个特征通过广义K-L变换降维并融合，产生的新特征吸收了单个特征的对模式分类的优势。同时, 分类器采用多神经网络模块结构，将以往BP网络规模大小与问题复杂性的矛盾转化为系统规模大小与问题复杂性的矛盾。利用遗传算法同时进行特征选择及网络结构优化，以构成有利于分类的自适应特征空间，在此分类子空间内，分别对各类训练样本进行选择优化，从而最终实现了一个自适应特征模块网络结构。手写体汉字的识别试验验证了所给方法的有效性。
4、通过对神经网络集成的理论分析，提出了一种多级神经网络结构的手写体汉字识别模型。第一级采用主分量分析神经网络，用于提取相关字符特征的主分量值，减小后级网络的运作规模。第二级为改进的GLVQ神经网络对PCA网络降维处理后的主分量特征对手写体汉字进行分类。GLVQ算法是从最优化一个目标函数而导出的，该算法构造新颖，为克服学习向量量化算法存在的问题提供了一个新的思路。分析了GLVQ算法的数学理论基础，完善了学习向量量化算法的理论，以及基于这些理论设计出更高效的分类器算法并把算法应用于有限集手写体汉字识别研究中。该算法不但可以有效提高系统的识别率，而且具有良好的泛化能力，这与传统的BP算法相比，有着明显的优势。
5、提出了一种多神经网络融合结构并将之应用于有限集手写体汉字识别。为了提高多神经网络分类器融合的效果，采用了一种改进的证据理论融合方法，通过对大量样本的统计，获得有关每个分类器识别性能的先验知识，将其作为证据合成的依据。针对证据融合中计算复杂问题，推导了一种快速算法。在有限集汉字识别系统中的实验结果表明，不同的特征和分类器从不同的角度刻画了手写体汉字图像的本质，充分利用这些特征和分类器提供的信息，作出一个更为客观的决策是可行的。
The information fusion technique has been widely and successfully applied to many fields, and is valuable especially to pattern recognition. It has been turned out that, for complicated pattern recognition problems such as recognizing handwritten characters, there exists no reliable and simple way to perform a high recognition rate. Every method has its advantages, limitations and applications, and is complementary for each other. So how to combine the advantages of various methods to develop an information fusion recognition system and overcome the disadvantages became one of the hottest themes of pattern recognition. The main works of this paper are listed as follows:
1. To handle the deformations is one of the key problems of handwritten Chinese characters recognition. The goal of normalization is to correct the deformation of handwritten character images and reduce the differences of the character images, which belong to the same class. A new nonlinear normalization method was developed, after I carefully researched the existed normalization methods. This method is based on line density method, and the line density filling in algorithm is used to perform the coordinate transformation. This makes the description of the stroke density more reasonable, and the distribution of strokes in normalized character image more proportional.
2. The features of Chinese characters characterize the unitary and local shape of the characters. Under a set of well-selected features, the distances of character samples of same class ought to be small; at the same time, the distances of character samples of different classes ought to be prominently large. According to the characteristics of handwritten Chinese characters, I developed some new features such as improved directional element feature, stroke sub-area moment, extended circumjacent stroke directional element feature, etc. These features are proved to be effective in experiments.
3. I studied the handwritten Chinese character recognition techniques based on adaptive information fusion and module neural networks. I provided a new method of multi-feature fusion on the basis of the
traditional feature-extraction methods. The high dimension of feature vector can be reduced by generalized K-L transformation, and the new feature vector is produce with the predominance of the single features for classification. The classifier is a multi-module neural network, so the conflict between the scale of BP network and the complexity of recognition tasks can be transformed to the conflict between the scale of recognition system and the complexity of recognition tasks. In the meantime, an adaptive feature subspace was constructed by using genetic algorithm to select features and optimize the structure of the neural network. The subspace of classification was optimized by training samples too. Finally, an adaptive feature-generating neural network module was formed. The recognition experiments have verified the method.
4. By analyzing the theory of neural network integration, I developed a multi-level neural network model for recognition handwritten Chinese characters. The PCA network is used as the first level to extract the principle component of the features of characters, and reduce the scale of the neural network of subsequent level. The second level is a GLVQ network. The function of the level is to classify the principle component feature vectors produced by the PCA network. The novel GLVQ algorithm was induced to optimize an object function, it provides a new way to overcome the exist problem of LVQ algorithms. I analyzed and improved the basic mathematic theory of GLVQ, designed a more efficient classification algorithm based on the theory, and applied the algorithm to handwritten Chinese character recognition. The algorithm improves the recognition rate evidently, and has well capability of generalization. It is obviously superior to traditional BP algorithms.
5. A multi-neural network fusion structure was developed and was applied to recognize the limited set of
引文
[1] 康耀红.数据融合及应用.西安：西安电子科技大学出版社，1997
[2] Swan J.,and Shields F.Multi- sensor fusion methodologies compared.SPIE,1991.1 483:219-229
[3] 阎福旺.信息融合技术. 北京：海洋出版社，2001
[4] 何友等.多传感器信息融合及应用.北京：电子工业出版社，2000
[5] 申功勋、孙建峰.信息融合理论在惯性/天文/GPS组合导航系统中的应用.北京：国防工业出版社，1998
[6] 滕召胜，罗隆福，童调生.智能检测系统与数据融合.北京：机械工业出版社，2000
[7] 郁文贤，雍少为，郭桂蓉.多传感器信息融合技术述评.国防科技大学学报.1994，16（3）：1-11
[8] Cao j,Shridhar M,and Ahmadi M.Handwritten numeral recognition with neural networks and information fusion.Midwest Symposium on Circuits and Systems, 1994,p 569-572
[9] 戴汝为，郝红卫，肖旭红.汉字识别的系统与集成.杭州：浙江科学技术出版社，1998
[10] 吴佑寿,丁晓青.汉字识别原理方法与实现.北京：高等教育出版社，1992
[11] 刘昌平，钱跃良，张永慧，宋东，李丰林.863手写汉字识别测试平台.中文信息 学报，1999,14(2):2-7
[12] Hildebrandt T.Y.,and Liu W..Optical recognition of handwritten Chinese characters: advance since 1980.Pattern Recognition,1993,26(2):205-225
[13] Mori S.,Suen C.Y.,and Yamamoto K..Historical review of OCR research and development. Proceedings of the IEEE,1992.80(7):1029-1058
[14] 张德喜.手写体汉字机器识别技术的现状分析.许昌师专学报.1999，18（3）：91-94
[15] Zhang Xinzhong.Feature point method of Chinese character recognition and its application. Journal of Computer Science and Technology,1990, 5 ( 4): 305-311
[16] Chiu Hung-Pin and Tseng Din-Chang.Novel stroke-based feature extraction for handwritten Chinese character recognition.Pattern Recognition,1999,32(12): 1947-1959
[17] Fan Kuo-Chin and Wu Wei-Hsien.Run-length-coding-based approach to stroke extraction of Chinese characters.Pattern Recognition, 2000,33 (11): 1881-1895
[18] 潘保昌.一种识别手写字符的新方法及系统实现.博士学位论文, 重庆大学，重庆，1990
[19] Ip Wilson W.S., Chung Korris, F.L Yeung and Daniel S.Offline handwritten Chinese character recognition via radical extraction and recognition. Proceedings of the International Conference on Document Analysis and Recognition, 1997, p 185-189
[20] Wang An-Bang, Fan Kuo-Chin,and Wu Wei-Hsien.Recursive hierarchical radical extraction for handwritten Chinese characters.Pattern Recognition,1997, 30( 7): 1213-1227
[21] Cheng-Lin Liu, In-Jung Kim and Jin H. Kim.Model-based stroke extraction and matching for handwritten Chinese character recognition, Pattern Recognition, 2001,34(12): 2339-2352
[22] 薛炳如,杨静宇,娄震等.小类别数手写汉字建模.中文信息学报,1999,13(6)：34-40
[23] 陈治平,林亚平,李军义.基于笔划和笔顺的汉字识别算法.湖南大学学报 (自然科学版 ),2000, 2 7(4)：102-105
[24] 孙星明,杨茂江,刘国华.完全基于结构知识的汉字笔画抽取方法.计算机研究与发展.2000,37 ( 5):533-549
[25] Yamamoto K. and Rosenfeld A. Recognition of hand-printed kanji character by a relaxation method. Proceedings - International Conference on Pattern Recognition, 1982 , p 395-398
[26] Sekita Iwao,Toraichi Kazuo,Mori Ryoichi,Yamamoto Kazuhiko,and Yamada Hiromitsu.Feature extraction of handwritten Japanse character by spline functions for relaxation matching. Pattern Recognition,1988, 21(1):9-17
[27] Leung Cheung-Hoi,Cheung Y. S.,and Wong Y. L..Knowledge-based stroke-maching method for chinese character recognition. IEEE Transactions on Systems, Man and Cybernetics, 1987, SMC-17( 6): 993-1003
[28] Xie S. L., and Suk Minsoo.On machine recognition of hand-printed Chinese character by feature relaxation.Pattern Recognition, 1988, 21(1 ):1-7
[29] Cheng Fang-Hsuan,Hsu Wen-Hsing,and Kuo Ming-Chuan.Recognition of handprinted Chinese characters via stroke relaxation. Pattern Recognition, 1993, 26 (4 ): 579-593
[30] Cheng Fang-Hsuan.Multi-stroke relaxation matching method for handwritten Chinese character recognition. Pattern Recognition,1998,31( 4 ):401-410
Chen Ling-Hwei Lieh,and June-Rong. Handwritten character recognition using a 2-layer random graph model by relaxation matching.Pattern Recognition
[31] Letters,1990 ,23(11):1189-1205
[32] Wang An-Bang,Fan, Kuo-Chin,and Huang, Jun S.Recognition of handwritten Chinese characters by modified relaxation methods. Image and Vision Computing, 1994, 12(8) : 509-522
[33] Huang Xiaofei,Gu Jun, and Wu Youshou.Constrained approach to multifont Chinese character recognition.IEEE Transactions on Pattern Analysis and Machine Intelligence. 1993,15(8) : 838-843
[34] Lippmann R.P..An introduction to computing with neural nets.IEEE Acoustics, Speech and Signal Processing Magzine,1987,4(2):4-22
[35] Wang Sheng-De,and Pan Chung-Chi.A neural network approach for Chinese character recognition. 1990 International Joint Conference on Neural Networks - IJCNN 90 Part 2 (of 3) , 1990,p 917-923
[36] Fu Alan M.,Yan Hong,and Suters Mark.Flexible pattern matching using a Hopfield-Amari neural network. Optical Engineering,1995,34(8):2467-2474
[37] Gan K.W.,and Lua, K.T.Chinese character classification using an adaptive resonance network. Pattern Recognition, 1992 , 25( 8) :877-882
[38] Jin Lianwen, Chan Kwokping,and Xu Bingzheng.Off-line Chinese handwriting recognition using multi-stage neural network architecture. IEEE International Conference on Neural Networks - Conference Proceedings 6, 1995 , p 3083-3088
[39] Chung Ho Sun, Ryeu, Jin Kyung,and Lee Wu Il.Input driven MLP model and its application to the recognition of Chinese characters. Proceedings - IEEE International Symposium on Circuits and Systems,1991,v 2, p 1192-1195
[40] 金连文，徐秉铮.RBF-DDA神经网络用于手写体汉字识别的研究.华南理工大学学报(自然科学版)，1999，25（9）：91-101
[41] Romero Richard D., Touretzky David S.,and Thibadeau Robert H.Optical Chinese character recognition using probabilistic neural networks. Pattern Recognition,1997 30 ( 8 ): 1279-1292
[42] Suganthan P.N.,and Yan H. Handwritten Chinese character recognition by ARG matching using self-organising Hopfield neural network.IEEE International Conference on Neural Networks - Conference Proceedings 3 ,1996, p 1456-1461
[43] 高涛，李明敬，李志峰.一种多字体特大字符集字符识别系统.中文信息学报，1999，14（2）：31-36
Rau Jen-Da,and Wang Jung-Hua.Voting principle of multiple features for Chinese character recognition system using neural network classifiers. Proceedings of the
[44] IEEE International Conference on Systems, Man and Cybernetics 6 ,1999 ,p VI-874 - VI-878
[45] Kovaca-V Zs M.Novel architecture for high quality hand-printed character recognition .Pattern Recognition, 1995, 28(11): 1685-1692
[46] Toraichi Kazuo, Kumamoto Tadahiko,Sekita Iwao, Yamamoto Kazuhiko,andYamada Hiromitsu.Feature analysis of handprinted Chinese characters for a recognition system. Proceedings - ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing,1989, 3, p 1715-1718
[47] 田盛丰，黄厚宽，李洪波.基于支持向量机的手写体相似字识别.中文信息学报，14（3）：37-41
[48] 张军，戚飞虎，叶芗芸.用于字符识别的椭圆基函数概率神经网络.红外与毫米波学报，1998，17（2）：91-97
[49] 徐涛，胡振宁，骆明.一个基于SDM改进模型的手写体汉字联想记忆系统的实现.微电子学与计算机，2000，1：1-4
[50] 彭健.小字符集自适应字符识别技术及系统的研究. 博士学位论文.重庆：重庆大学，2002
[51] 张洪刚.HCL2000汉字数据库的建立与高精度识别算法的研究.研究生学位论文.北京：北京邮电大学，1999.3
[52] Mori Shunji,Yamamoto Kazuhiko,and Yasuda Michio.Research on machine recognition of handprinted characters. IEEE Transactions on Pattern Analysis and Machine Intelligence, 1984, PAMI-6(4 ):386-405
[53] Otsu Nobuyuki.Threshold selection method from gray-level histograms.IEEE Transactions on Systems, Man and Cybernetics,1979,SMC-9(1) : 62-66
[54] 叶芗芸,戚飞虎,吴健渊.文本图像的快速二值化方法.红外与毫米波学报,1999,16(5):344-350
[55] Yamasaki Issei,and Iijima Taizo.Basic equation of automatic normalizer form character patterns using moments. Electronics and Communications in Japan,1972,55(5): 148-156
[56] Yamashita Y,Higuchi K,and Yamada Y,etal.Classification of handprinted Kanji characters by the structured segment matching method. Pattern Recognition Letters, 1983,1:475-479
[57] Yamada Hiromitsu.Nonlinear methods in character recognition. Researches of the Electrotechnical Laboratory,1985,853 :0366-9106
Tsukumo Jun ,and Tanaka Haruhiko.Classification of handprinted Chinese characters
[58] using non-linear normalization and correlation methods.Proceedings - International Conference on Pattern Recognition, Nov 14-17 1988,p 168-171
[59] Yamada Hiromitsu,Yamamoto Kazuhiko,and Saito Taiichi. Nonlinear normalization method for handprinted Kanji character recognition - line density equalization. Proceedings - International Conference on Pattern Recognition. Nov 14-17 1988, p 172-175
[60] Lee S.-W.,Park, J.-S.,and Tang, Y.Y..Performance evaluation of nonlinear shape normalization methods for the recognition of large-set handwritten characters.Proceedings of the 2nd International Conference on Document Analysis and Recognition, Oct 20-22 1993 , p 402
[61] Lee Seong-Whan, and Park Jeong-Seon.Nonlinear shape normalization methods for the recognition of large-set handwritten characters. Pattern Recognition,1994,27(7): 895-902
[62] Chen Youbin, Ding Xiaoqing,Wu Youshou, and Chen Ming.New nonlinear shape normalization method for off-line handwritten Chinese character recognition , Lecture Notes in Computer Science,1998, 1351 p I-168
[63] 陈友斌，丁晓青，吴佑寿.一种新的用于手写汉字识别的非线性规一化方法.模式识别与人工智能，1998，11（3）：310-317
[64] 王庆，赵荣椿，冯大淦.手写体汉字的规范化处理及评价.数据采集与处理，2001，16(2)：227-232
[65] Wakahara Toru ,and Odaka Kazumi.Adaptive normalization of handwritten characters using global/local affine transformation. Proceedings of the International Conference on Document Analysis and Recognition, Aug 18-20 1997 v1, p 28-33
[66] Kim Sang-Yup, Lee Seong-Whan.Nonlinear shape normalization methods for gray-scale handwritten character recognition.Proceedings of the International Conference on Document Analysis and Recognition, Aug 18-20 1997, v2, p 479-782
[67] Hamanaka Masahiko,Yamada Keiji,and Tsukumo Jun.Handprinted Kanji character recognition using normalization-cooperative feature extraction method. NEC Research & Development, 1994,35(3) : 327-336
[68] Liu Cheng-Lin,Kim In-Jung, Kim Jin H.High accuracy handwritten Chinese character recognition by improved feature matching method.Proceedings of the International Conference on Document Analysis and Recognition,1997,2,p 1033-1037
[69] Yamada Hiromitsu.Continuous nonlinearity in character recognition .IEICE Transactions on Information and Systems, 1996,E79-D5:423-428
[70] Wang Song, Shu ChangXia,and Shaowei.Hybrid approach to unconstrained handwritten numerals recognition. International Conference on Signal Processing Proceedings, Oct 14-18 1996,v2 p 1334-1337
[71] Horiuchi Takahiko,Haruki Ryoji,Yamada Hiromitsu,and Yamamoto Kazuhiko. Two-dimensional extension of nonlinear normalization method using line density for character recognition. Proceedings of the International Conference on Document Analysis and Recognition, Aug 18-20 1997,v 2 p 511-51
[72] Der-Sheng Lin,and Jin-Jang Leou.A Genetic Algorithm Approach to Chinese Handwriting Normalization.IEEE Transactions on systems,man,and cybernetics -part B:cybernetics,1997,27(6) :999-1006
[73] Horiuchi Takahiko,Haruki Ryoji,Yamada Hiromitsu,and Yamamoto Kazuhiko. Two-dimensional extension of nonlinear normalization method using line density. IEICE Transactions on Information and Systems, 1997,J80-D-II(6):1600-1607(in Japanese)
[74] Horiuchi Takahiko. Two-dimensional nonlinear normalization by transforming shape pixels. IEICE Transactions on Information and Systems, 1999, J82-D-II(12):2428-2431(in Japanese)
[75] Y.H. Tseng, C.C. Kuo, H.J. Lee.Speeding up Chinese character recognition in an automatic document reading system. Pattern Recognition ,1998,31 (11):1601-1612
[76] 孙即祥等.现代模式识别. 长沙：国防科技大学出版社，2002
[77] 边肇祺，张学工等.模式识别（第二版）.北京：清华大学出版社，2000
[78] Chen Youbin, Ding Xiaoqing,and Wu Youshou.Off-line handwritten Chinese character recognition based on crossing line feature.Proceedings of the International Conference on Document Analysis and Recognition, Aug 18-20 1997, v1, p 206-210
[79] R. Oka. Handprinted Chinese-Japanese character recognition by using cellular feature. Proc. 6th ICPR , 1982,p. 783-785
[80] M. Umeda.Advances in recognition methods for handwritten Kanji characters. IEICE Trans. Inf.&Syst.,1996,E79-D:401-410
[81] Oivind Due Trier,Jain A K,and Torfinn Tax.Feature extraction method for character recognition-a survey.Pattern Recognition,1996,29(4):662
Kimura Fumitaka ,Wakabayashi Tetsushi, Tsuruoka Shinji,and MiyakeYasuji.
[82] Improvement of handwritten Japanese character recognition using weighted direction code histogram. Pattern Recognition,1997,30(8):1329-1337
[83] S. Tsuruoka, M. Kurita, T.Harada, F. Kimura, and Y. Miyake.Handwritten "Kanji" and "Hiragana"character recognition using weighted direction index histogram method. IECE Trans. 1987,J70-D: 1390-1397, (in Japanese).
[84] T. Akiyama ,and N. Hagita.Automated entry system for printed docments. Patten Recognition ,1990,23:1141-1154
[85] S. Akamatsu ,and T. Kawatani. Hierarchical classication of handprinted Kanji using density feature and conguration feature, Proc. 6-th IJCPR,1982 , p. 783
[86] 熊毅之，卞松玲，刘明，陶纯堪. 平移、旋转及尺度不变光学神经网络识别系统. 红外与激光工程，1997，05:17-21
[87] 金连文，彭秀兰，尹俊勋.一种手写体汉字特征提取新方法--小波变换及弹性网格技术的应用.中国图像图形学报. 1998,3(7)：549-552.
[88] 尹俊勋,张青. 一种新的手写体汉字笔划提取方法. 华南理工大学学报(自然科学版).1998, 26(9):73-77
[89] 孙立民,狄红卫,余英林.基于子块特征及其相关模糊特征的手写体汉字识别方法.通信学报.1999,2 0(12):81-85
[90] 马少平,夏 莹,朱小燕. 基于模糊方向线素特征的手写体汉字识别. 清华大学学报(自然科学版). 1997,37(3):42-45
[91] 李建平. 小波分析与信号处理. 重庆：重庆出版社. 1997
[92] 程正兴. 小波分析算法与应用. 西安：西安交通大学出版社，1997
[93] Mallat S. A theory for multi-resolution signal decomposition: The wavelet representation.IEEE Trans. Pattern Anal Mach Intell. 1989. 11(7): 674-693
[94] 叶俊勇.人脸检测与识别方法研究.博士学位论文，重庆大学，重庆 2002
[95] Leung C.H.,and SZE L..Feature selection in the recognition of handwritten Chinese characters.Engng Applic. Intell.,1997,10(5):495-502
[96] N. Sun, M. Abe, and Y. Nemoto.A handwritten character recogniton system by using improved directional element feature and subspace method. IEICE Trans. ,1995,J78-D-II: 922-930 (in Japanese)
[97] Y. Mizukami.A handwritten Chinese character recognition system using hierarchical displacement extraction based on directional features. Pattern Recognition Letters , 1998 ,19 :595-604
Kato Nei, Suzuki Masato, Omachi Shin'ichiro ,Aso Hirotomo,and Nemoto, Yoshiaki.Handwritten character recognition system using directional element feature
[98] and asymmetric Mahalanobis distance. IEEE Transactions on Pattern Analysis and Machine Intelligence ,1999, 21 (3): 258-262
[99] Chim Y. C, Kassim A. A, Ibrahim Y.Character recognition using statistical moments.Image and Vision Computing. 1999, 17 (3-4):299-307
[100] Liao Simon X,Lu Qin.Study of moment functions and its use in Chinese character recognition.Proceedings of the International Conference on Document Analysis and Recognition.ICDAR, IEEE, Aug 18-20 1997,2 : 572-575
[101] Dehghan Mehdi,Faez, Karim. Farsi handwritten character recognition with moment invariants.International Conference on Digital Signal Processing, DSP, IEEE, IEE,EURASIP IEEE,Jul 2-4 1997 ,2 :507-510
[102] Tsang I. J., Tsang I. R,Van Dyck D.Handwritten character recognition based on moment features derived from image partition .IEEE International Conference on Image Processing, Oct 4-7 1998,2:939-942
[103] 居琰,汪同庆,刘建胜,王贵新,袁祥辉.基于笔划分区矩特征的有限集手写体汉字识别.计算机工程，2003, 1.待刊
[104] N. Hagita, S. Naito, and I. Masuda.Handprinted Chinese characters recognition by peripheral direction contributivity feature. IECE Trans., 1983,J66-D: 1185-1192 (in Japanese)
[105] H. Sakano ,and N. Miyamoto.Handprinted character recognition by extended peripheral direction contributivity.Proc. 1995 IEICE General Conference , 1995 , p. 282 (in Japanese)
[106] Kimura Yoshimasa, Akiyama Teruo, Mori Minoru,and Miyamoto Nobuo, et al.Handprinted Kanji character recognition software using extended peripheral direction contributivity features.NTT R&D,1998, 47(6): 617-622 (In Japanese)
[107] Minoru Mori, and Toru Wakahara.Handwritten Kanji Character Recognition Using Relative Direction Contributivity. IEICE Transaction of Information and Systems（D-II）, 2001,J84-D-II(7):1360-1368 (in Japanese)
[108] Minoru Mori, Toru Wakahara, and Kenji Ogura.Measures for Structural and Global Shape Description in Handwritten Kanji Character Recognition. SPIE/IS&T's Symposium on Electronic Imaging '98 Document Recognition V, 1998 vol.3305, p.81-89
[109] 阎平凡，张长水.人工神经网络及模拟进化计算.北京：清华大学出版社，2000
[110] 侯平魁，史习智,林良骥,王学军.水下目标识别的特征融合分类器设计.电 子学报，2001，29（4）：443-446
[111] 刘永红.神经网络理论的发展与前沿问题.信息与控制，1999，28（1）：31-44
[112] J T Favata,G Srikantan, and S N Srihari..Handprinted character/digit recognition using a multiple feature/resolution philosophy. Proc.IWPHR'94, 1994:57-66
[113] T Matsuii,T Tsutsunida,and S N Srihari.Combination of stroke/background structure and contour-direction features in handprinted alphanumeric recognition.Proc. of IWFHR'94,1994:87-96
[114] Kimura Fumitaka,Nishikawa Shuji,and Wakabayashi Tetsushi,et al.Evaluation and synthesis of feature vectors for handwritten numeral recognition.IEICE Trans.INF.&SYST.,1996, E79-D(5):436-442
[115] 李在铭等.数字图像处理压缩与识别技术.成都：电子科技大学出版社，2000
[116] 章毓晋.图像处理和分析.北京：清华大学出版社，1999
[117] Yang Jian., Yang Jing-yu.Generalized K-L transform based combined feature extraction.Pattern Recognition, 2002,35: 295-297
[118] 戴华.矩阵论.北京：科学出版社，2001
[119] 《现代应用数学手册》编委会.现代应用数学手册：概率统计与随机过程卷，北京清华大学出版社，2000
[120] Oh Il-Seok, Suen Ching Y.. Feature for character recognition based on directional distance distributions.Proceedings of the International Conference on Document Analysis and Recognition, ICDAR, Aug 18-20 1997, v 1, p 288-292
[121] 乔进.自由手写体字符脱机识别方法的研究.博士学位论文，重庆大学，重庆 1999.6
[122] John G.,Kohavi R.,and Pfleger K..Irrelevant features and the subset selection problem.Proceedings of the Eleventh International Conference on Machine Learning,1994,p.121-129
[123] Narendra P. M.,and Fukunaga K..A branch and bound algorithm for feature subset selection.IEEE Trans.Comput.,1977,26(9):917-922
[124] Glover F..Tabu search 1.ORSA J.on Computing,1989,1:190-206
[125] 张鸿宾，孙广煜.Tabu搜索在特征选择中的应用.自动化学报，1999，25（4）：457-466
[126] 岳喜才，伍晓字，郑崇勋，叶大田.一种大类别数分类的神经网络方法.计算机研究与发展，2000，37（3）：278-283
[127] R.A. Jacobs, and M.I. Jordan.A competitive modular connectionist architecture. in Advances in Neural Information Processing Systems ,1991,Vol.3, p.767-773
T. Hrycej. Modular Learning in Neural Networks: A Modularized Approach to
[128] Neural Network Classification.John Wiley & Sons, New York, 1992
[129] E.L. Schwartz (Ed.). Computational Neuroscience. MIT Press, Cambridge, 1990
[130] D.H. Ballard.Modular learning in hierarchical neural networks. in Computational Neuroscience (Edited by E.L. Schwartz), p.139-153, MIT Press, Cambridge, 1990
[131] S.-H. Chen ,and Y.-F. Liao.Modular recurrent neural networks for Mandarin syllable recognition. IEEE Tr. on Neural Networks, , 1998, 9(.6):1430-1441
[132] A. Waibel, H. Sawai, and K. Shikano.Modularity and scaling in large phonemic neural networks. IEEE Tr. on Acoustics, Speech, and Signal Processing, 1989, 37, (12):1888-1898
[133] S.-C. Tsay, P.-R. Hong, and B.-C. Chieu.Handwritten digits recognition system via OCON neural network by pruning selective update. Proceedings of 11th ICPR, 1992,p.656-659
[134] Mui, A. Agarwal, A. Gupta, P.S.P. Wang.An adaptive modular neural network with application to unconstrained character recognition. in Document Image Analysis,1994,p.1189-1203
[135] LR. Anand, K. Mehrotra, C.K. Mohan, and S. Ranka.Efficient classification for multiclass problems using modular neural networks. IEEE Tr. on Neural Networks, 1995, 6(1):117-124
[136] I.-S. Oh, J.-S. Lee, K.-C. Hong, and S.-M. Choi.Class-expert approach to handwritten numeral recognition. Proceeding of 5th IWFHR, 1996, pp.35-40
[137] 郑文波，林雅梅等.无限制性手写数字的多神经网络识别. 福州大学学报 (自然科学版)，1 998， 2 6（2）：18-21
[138] 王 伟，盛立东.基于级连分组BP网络的高精度手写数字识别.中文信息学报，1999，1 4（2）：60-64
[139] 宋红萍，刘宏超，全安寿，崔晋川，汤映杰.基于小波网络和多模块网络的数字识别.中文信息学报，1999，1 4（2）：37-42
[140] Il-Seok Oh ,and Ching Y. Suen.A class-modular feedforward neural network for handwriting recognition.Pattern Recognition, 2002,35 (1):229-244
[141] Il-Seok Oh, Jin-Seon Lee, and Ching Y. Suen.Analysis of class separation and combination of class-dependent features for handwriting recognition. IEEE Tr. on PAMI, 1999, 21(10)1089-1094
[142] [美] Abhijit S，Pandya Robert，and B. Macy.神经网络模式识别及其实现.徐勇,荆涛等译.北京：电子工业出版社，1999
Il-Seok Oh, Jin-Seon Lee, and C.Y. Suen.A class modularity for character
[143] recognition. Proceedings of 6th International Conference on Document Analysis and Recognition, 2001, p.64-68
[144] Holland J H. Adaptation in Nature and Artificial Systems. MIT Press,1992
[145] De Jong K A. An Analysis of the Behavior of a class of Genetic Adaptive Systems .Ph.D Dissertation, University of Michigan,No.76-9381,1975
[146] Goldberg D E. Genetic Algorithms in Search, Optimization and Machine Learning. Addison -Wesley.1989
[147] 朱剑英.智能系统非经典数学方法.武汉：华中科技大学出版社，2001
[148] 靳蕃.神经计算智能基础原理方法.成都：西南交通大学出版社，2000
[149] 黎明，严超华，刘高航.基于遗传策略和神经网络的非监督分类方法.软件学 报，1999，10（12）：1310-1315
[150] Yang, J.， and Honavar, V. (1998). Feature Subset Selection Using a Genetic Algorithm. IEEE Intelligent Systems (Special Issue on Feature Transformation and Subset Selection)，13:44-49.
[151] Daming Shi, Wenhao Shu, and Haitao Liu.Feature selection for handwritten Chinese character recognition based on genetic algorithms, SMC'98 Conference Proceedings ,1998, vol.5, p.4201-4206
[152] J. Moody,and J. Utans.Principled architecture selection for neural networks:Application to corporatebond rating preiction,Advances in Neural Information Processing Systems 4,Morgan Kaufmann,1991
[153] Hornik K M,Stinchcombe M,White H. Multilayer feed forward networks are universal approximators.NeuralNetworks,1989,2(2):359-366
[154] 阎平凡.对多层前向神经网络研究的进一步看法.电子学报，1999，27（5）：82-85
[155] Prager Richard W.CART/CMAC hybrid: Regression trees with interpolation . Proceedings - International Conference on Pattern Recognition ,1994 , p 476-478
[156] Psichogios Dimitris C.,De Veaux Richard D,and Ungar Lyle H.. Non- parametric system identification: A comparison of MARS and neural networks. Proceedings of the American Control Conference ,1992 ,v 2, p 1436-1440
[157] Yun-tao Qian,and Wei-xin Xie.Building fuzzy neural classifiers by fuzzy ID3 algorithm ,International Conference on Signal Processing Proceedings, ICSP,1996, v 2 , p 1292-1295
Chi Zheru,Suters Mark,and Yan Hong.Handwritten digit recognition using combined ID-3-derived fuzzy rules and Markov chains.Pattern Recognition,
[158] 1996,29(11): 1821-1833
[159] 周志华，陈世福.神经网络集成.计算机学报，2002，25（1）：1-8
[160] Hansen L K,SalamonP.Neural network ensembles.IEEE Trans Pattern Analysis and Machine Intelligence,1990,12(10):993-1001
[161] Sollich P,Krogh A.Learning with ensembles:Howover-fitting can be useful.In:Touretzky D,Mozer M,Hasselmo Meds. Advances in Neural Information Processing Systems8,Cambridge,MA:MITPress,1996.190-196
[162] Kearns M,Valiant L G. Learning Boolean formulae or factoring. Aiken Computation Laboratory,Harvard University,Cambridge,MA,Technical Report:TR-1488,1988
[163] Breiman Leo.Bagging predictors.Machine Learning ,1996, 24( 2):123-140
[164] Schapire R E. The strength of weak learnability. Machine Learning, 1990,5(2):197-227
[165] Freund Y. Boosting a weak algorithm by majority. Information and Computation,1995,121(2):256-285
[166] Freund Y., Schapire R.E..A Decision-theoretic generalization of on-line learning and an application to boosting Journal of Computer and System Sciences,1997,55 (1 ): 119-139
[167] Robert E. Schapire, Yoav Freund, et al.Boosting the Margin: A New Explanation for the Effectiveness of Voting Methods Annals of Statistics, 1998,26(5):1651-1686
[168] Mao Jianchang.Case study on Bagging, Boosting, and Basic ensembles of neural networks for OCR. IEEE International Conference on Neural Networks - Conference Proceedings,1998, v 3 ,p 1828-1833
[169] Mao Jianchang,and Mohiuddin K.M.Improving OCR performance using character degradation models and boosting algorithm. Pattern Recognition Letters,1997, 18 (11-13):1415-1419
[170] Suganthan P.N.Hierarchical overlapped SOM's for pattern classification.IEEE Transactions on Neural Networks,1999, 10(1) :193-196
[171] Suganthan P.N.Pattern classification using multiple hierarchical overlapped self-organising maps,Pattern Recognition,2001,34(11):2173-2179
[172] Jian-xiong Dong, Krzyzak A, Suen C. Y. , Local learning framework for recognition of lowercase handwritten character. International workshop on Machine Learning and Data Mining in Pattern Recognition, 2001, p. 226-238
[173] Jian-xiong Dong, Krzyzak A, Suen C. Y. , A Local Learning framework for Pattern Classification. Proceedings of the 14th international conference on Vision Interface., 2001, p. 220-227
[174] Kazuki Saruta, Nei Kato, Masato Abe, Yoshiaki Nemoto. High Accuracy Recognition of ETL9B Using Exclusive Learning Neural Network-Ⅱ( ELNET-Ⅱ), 1996,E79-D(5):516-512
[175] Ahmad Fadzil Abif, Hidekazu Takahashi, Akira Iwata,and Toshio Tsutsumida.Handwritten Postal Code Recognition by Neural Network --A Comparative Study,1996,E79-D(5):443-455
[176] R. Vigario ,and E. Oja.Signal Separation and Feature Extraction by Nonlinear PCA Network. Proc. 9th Scandinavian Conference on Image Analysis, 1995,p. 811-818.
[177] J. Rubner,and P. Tavan.A self-organizing network for pricipal component analysis. Europhysics Lett.,1989,10:693-698
[178] 张军.神经网络在模式识别中的应用研究.博士学位论文.上海：上海交通大学,1997
[179] Gray R M.Vector Quantization.IEEE Assp Magazine,1984,1:4-29
[180] Kohonen T.The self-organizing map.Proc IEEE,1990,78:1464-1480
[181] Kohonen T.Improved versions of learning vector quantizations.Int. Joint Conf. Nrural Networks,I,1990,p545-550
[182] Sato A.,and Yamada K..Generalized learning vector quantization.Advances in Neural Information Processing 8, Proceedings of the 1995 Conference, 1996,p423-429
[183] Tsay Mu-King,Shyu Keh-Hwa,and Chang Pao-Chung.Feature transformation with generalized learning vector quantization for handwritten chinese character recognition.IEICE Trans. INF.&SYST.,1999,E82-D(3):687-692
[184] 张志华,郑南宁,王天树.广义LVQ神经网络的性能分析及其改进.自动化学报,1999,25(5):583-589
[185] Juang B.H.,and Katagiri S.Discriminative learning for minimum error classification.IEEE Trans. Signal Processing,1992,40(2):3043-3054
[186] Sato a.,and Yamada K..A formulation of learning vector quantization using a new misclassification measure.Proc. of ICPR'98,1998,p332-325
[187] M. Ikeda,H. Tanaka,and T. Moto-oka.Projection distance method for recognition of handwritten characters.Trans.IPS Japan,1983,24(1):106-112(in Japanese)
[188] Oja e..Subspace method of pattern recognition.Reserch Studies Press,England,1983
[189] Tsuyoshi Fukumoto, Tetsushi Wakabayashi, Fumitaka Kimura, and Yasuji Miyake.Accuracy Improvement of Handwritten Character Recognition by GLVQ.7th International Workshop on Frontiers in Handwriting Recognition Proceedings, 2000,p.271-280
[190] 胡钟山,娄震,杨静宇,刘克,孙靖夷.基于多分类器组合的手写体数字识别.计算机学报.1999,22(4):369-374
[191] Kawatan, Takahiko,and Shimizu Hiroyuki.Complementary classifier design using difference principal components.Proceedings of the International Conference on Document Analysis and Recognition, ICDAR, 1997, v 2 , p 875-880
[192] Wang Jianguo, Yan Hong.Hybrid method for unconstrained handwritten numeral recognition by combining structural and neural `gas' classifiers.Pattern Recognition Letters, 2000,21(6) :625-635
[193] 刘汝杰,袁保宗,唐晓芳.一种新的基于聚类的多分类器融合算法.计算机研究与发展,2001,38(10)：1236-1240
[194] Xu Lei, Krzyzak Adam,and Suen Ching Y.Methods of combining multiple classifiers and their applications to handwriting recognition. IEEE Transactions on Systems, Man and Cybernetics,1992, 22(3): 418-435
[195] Lam Louisa ,and Suen Ching Y.Application of majority voting to pattern recognition: An analysis of its behavior and performance.IEEE Transactions on Systems, Man, and Cybernetics Part A:Systems and Humans, 1997 ,27 (5): 553-568
[196] Lam Louisa,and Suen Ching Y.Optimal combinations of pattern classifiers Pattern Recognition Letters,199516(9): 945-954
[197] Suen C. Y., Nadal C., Mai T. A., Legault R., and Lam L..Computer recognition of unconstrained handwritten numerals. Proc. IEEE, 1992, 80,p. 1162-1180
[198] Kang Hee-Joong,and Lee Seong-Whan.Dependency-based framework of combining multiple experts for the recognition of unconstrained handwritten numerals. Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition, 1999,v 2, p124-129
[199] Kang Hee-Joong,and Kim, Jin H.Probabilistic framework for combining multiple classifiers at abstract level .Proceedings of the International Conference on Document Analysis and Recognition, ICDAR,1997,v 2 , p870-874
Ho T.K., Hull, J., andJ. Srihari, S.N.. Decision combinationin multiple classifier
[200] systems. IEEE Transaction on Pattern Analysis and Machine Intelligence, 1994,16 (1): 66-75
[201] Munro P.W., Parmanto B.. Competition among networks improves committee performance. Advances in NeuralInformation Processing Systems 9, 1997， MIT Press, Cambridge,p592-598
[202] Jacobs R.A.. Methods for combining experts probability assessments.Neural Computation,1995,7(5):867-888.
[203] Chen W., Gader P.D.,and Shi H.. Improved dynamic programming-based handwritten word recognition using optimal order statistics. In: Proc. of the Internat. Conf. Statistical and Stochastic Methods in Image Processing II, 1997,p. 246-256
[204] Huang Y.S.,and Suen C.Y.Method of combining multiple classifiers - a neural network approach,Proceedings - International Conference on Pattern Recognition, 1994,2,p 473-475
[205] 吕岳，施鹏飞，赵宇明.改进的贝叶斯多分类器组合规则.数据采集与处理，2000，15(2):204-207
[206] Wesolkowski Slawomir, Hassanein Khaled.Comparative study of combination schemes for an ensemble of digit recognition neural networks .Proceedings of the IEEE International Conference on Systems, Man and Cybernetics ,1997,v 4,p 3534-3539
[207] 孙怀江,胡钟山,杨静宇.基于证据理论的多分类器融合方法研究.计算机学报,2001，24（3）：231-235
[208] 刘志言，童树鸿，王艳.基于证据理论的多分类器集成方法研究.电机与控制学报，2001，5（3）：208-212
[209] 叶芗芸,戚飞虎,朱国霞.一种多级分类器集成的字符识别方法.电子学报,1998,11：15-19
[210] 韩宏,杨静宇.神经网络分类器的组合.计算机研究与发展,2000, 37(12):1488-1492
[211] Verikas A,Lipnickas A,and Malmqvist K etal. Soft combinationof neural classifiers: A comparative study. Pattern RecognitionLetters,1999,20 (4):429- 444
[212] 肖旭红,戴汝为.一种识别手写汉字的多分类器集成方法.自动化学报, 1997,23(5)：621-627
[213] 段新生.人工智能-证据理论与决策.北京：中国人民大学出版社，1994
[214] Shafer G. A Mathematical Theory of Evidence. Princeton:Princeton University Press,1976
[215] Rogova G.. Combining the results of several neural network classifiers. Neural Networks, 1994, 7 (5): 777-781.