基于二阶系统的脱机中文签名鉴定的研究
|
摘要
签名是一种使用广泛的用于鉴定身份的生物特征,是社会生活普遍接受的一种同意或授权的方式。在电子商务银行业务、处理单据、签订合同等领域有广泛的应用,并具有相应的法律效力。但由于它的简单、易模仿性,也成为了伪造的目标,此时手写签名鉴定就会发挥它的作用。如果鉴定正确率高,将会对社会各行业包括行政、金融、处理法律纠纷、安全领域等方面能起到关键作用,因此在国家的经济建设中发挥重要作用。签名鉴定也是目前计算机模式识别领域的前沿课题。鉴于签名鉴定具有良好的应用前景和巨大的商业价值,世界各国许多学者和研究机构都对其表现出了极大兴趣。因此对签名技术的深入研究,对于提高实际应用水平和有关学科的科技进步具有着重要的意义。
本文针对脱机手写签名鉴定的实玑,首先介绍了脱机手写签名鉴定的发展现状及国内外研究动态,探讨了脱机中文签名鉴定实现的主要方法。然后深入分析和研究了脱机签名鉴定中的预处理技术、特征提取技术和鉴定技术,并进行了相关的实验。
在原始特征的提取时,为了尽可能体现签名的个体特征,同时提取了静态形状特征和伪动态特征。其中静态特征提取了几何特征和基于Zemike矩的形状特征;伪动态特征方面,我们提取了高灰度特征HPF、骨架方向灰度特征和灰度重心特征等。
由于书写风格不同,每个人的签名都有自己的特点,对于随意的模仿,使用形状特征进行鉴定就可以达到较高的正确鉴定率,但对精心模仿的签名就很难鉴别出真伪。因此本文提出了一种将静态特征与伪动态特征相融合的脱机中文签名鉴定系统。因为压力,速度等动态特征很难模仿,即使表面上可以以假乱真的签名其灰度还是有很大差别的。这种采用静态特征与灰度特征相结合进行辨识的技术,可提高系统鉴别真伪的能力。
另外,本论文尝试用2-阶鉴定的方法来提高鉴定的速度。这个二阶系统由两个独立的签名验证系统所构成。其中,第一阶段使用签名样本的几何特征和伪动态特征来鉴定签名的真伪,第二阶段则利用具有旋转不变的Zernike矩特征来鉴定签名的真伪。参考样本建立之后,当一个测试样本进入系统时,先以第一阶段的验证系统检查它。如果不符合我们的要求则可以直接判断该签名为伪签名,而不必进行下一阶段的鉴定;若通过第一阶段,则进入第二阶段,以Zernike矩特征来检查它的真伪。从实验中我们也证明了这一点是可行的,可以缩短简单伪造签名的鉴别时间,在不降低鉴定率的情况下我们提高了鉴定的速度。
Signature is one of the most common biological features used for identity verification.Because of the significant advantages of signature identification as a way of personal identification such as:universality uniqueness stability and so on, it has a wide perspective applications in electronic-bank,military affairs, electronic-business,communication,office automation engineering files and so on. The research on this topic is not only important for practical applications but also the development of science.
First,the application background,the development history and the researches at home and abroad of handwritten signature identification techniques are briefly introduced.The difficulties and main research methods of the off-line Chinese signature identification problem are analyzed.Then,the key techniques including image preprocessing,feature extraction and identification of off-line signature have been studied deeply in this thesis.The main researches include:
The signatures,static shape feature and pseudo dynamic features have been extracted in feature extraction.The emphasis of shape feature is the pseudo-Zernike moment with rotation,scale and translation invariance and anti-noise character.The pseudo dynamic features high gray feature(HPF),framework aspect gray feature, and gray barycenter feature.
In this thesis an off-line Chinese signature identification system based on pseudo feature and shape feature extraction is built furthermore,we adopted 2-pass identification method.As the writing styles are different,The signatures of different people vary from each other.The optional copy is easily identified using shape features.But it is difficult to find the intentional copies only using shape features. Imitating the writing press,peed and other dynamic features is difficult leads to superficially similar signatures have much difference.So we combine shape features with pseudo features to verify the using the feature fusion technique By this way we improve the ability of the signature identifications system.
We try to improve the speed by using 2-pass verifying.The identification of 2-pass system is composed of two independent signature identification system.The first phase uses simple geometric features and pseudo-dynamic features to identify the authenticity of signatures,while the second phase uses Zernike moment to make a confirmation After the establishment of reference samples,when one pattern enters, the first phase of the inspection system to verify it.If it does not meet our requirements,we can judge it is false;if it have adopted the first phase,we use the second phase,Zernike moment to examine the characteristics.It has been proved that the identification of 2-pass system can save a lot of time for identification with out descending the correct identification rate.
本文针对脱机手写签名鉴定的实玑,首先介绍了脱机手写签名鉴定的发展现状及国内外研究动态,探讨了脱机中文签名鉴定实现的主要方法。然后深入分析和研究了脱机签名鉴定中的预处理技术、特征提取技术和鉴定技术,并进行了相关的实验。
在原始特征的提取时,为了尽可能体现签名的个体特征,同时提取了静态形状特征和伪动态特征。其中静态特征提取了几何特征和基于Zemike矩的形状特征;伪动态特征方面,我们提取了高灰度特征HPF、骨架方向灰度特征和灰度重心特征等。
由于书写风格不同,每个人的签名都有自己的特点,对于随意的模仿,使用形状特征进行鉴定就可以达到较高的正确鉴定率,但对精心模仿的签名就很难鉴别出真伪。因此本文提出了一种将静态特征与伪动态特征相融合的脱机中文签名鉴定系统。因为压力,速度等动态特征很难模仿,即使表面上可以以假乱真的签名其灰度还是有很大差别的。这种采用静态特征与灰度特征相结合进行辨识的技术,可提高系统鉴别真伪的能力。
另外,本论文尝试用2-阶鉴定的方法来提高鉴定的速度。这个二阶系统由两个独立的签名验证系统所构成。其中,第一阶段使用签名样本的几何特征和伪动态特征来鉴定签名的真伪,第二阶段则利用具有旋转不变的Zernike矩特征来鉴定签名的真伪。参考样本建立之后,当一个测试样本进入系统时,先以第一阶段的验证系统检查它。如果不符合我们的要求则可以直接判断该签名为伪签名,而不必进行下一阶段的鉴定;若通过第一阶段,则进入第二阶段,以Zernike矩特征来检查它的真伪。从实验中我们也证明了这一点是可行的,可以缩短简单伪造签名的鉴别时间,在不降低鉴定率的情况下我们提高了鉴定的速度。
Signature is one of the most common biological features used for identity verification.Because of the significant advantages of signature identification as a way of personal identification such as:universality uniqueness stability and so on, it has a wide perspective applications in electronic-bank,military affairs, electronic-business,communication,office automation engineering files and so on. The research on this topic is not only important for practical applications but also the development of science.
First,the application background,the development history and the researches at home and abroad of handwritten signature identification techniques are briefly introduced.The difficulties and main research methods of the off-line Chinese signature identification problem are analyzed.Then,the key techniques including image preprocessing,feature extraction and identification of off-line signature have been studied deeply in this thesis.The main researches include:
The signatures,static shape feature and pseudo dynamic features have been extracted in feature extraction.The emphasis of shape feature is the pseudo-Zernike moment with rotation,scale and translation invariance and anti-noise character.The pseudo dynamic features high gray feature(HPF),framework aspect gray feature, and gray barycenter feature.
In this thesis an off-line Chinese signature identification system based on pseudo feature and shape feature extraction is built furthermore,we adopted 2-pass identification method.As the writing styles are different,The signatures of different people vary from each other.The optional copy is easily identified using shape features.But it is difficult to find the intentional copies only using shape features. Imitating the writing press,peed and other dynamic features is difficult leads to superficially similar signatures have much difference.So we combine shape features with pseudo features to verify the using the feature fusion technique By this way we improve the ability of the signature identifications system.
We try to improve the speed by using 2-pass verifying.The identification of 2-pass system is composed of two independent signature identification system.The first phase uses simple geometric features and pseudo-dynamic features to identify the authenticity of signatures,while the second phase uses Zernike moment to make a confirmation After the establishment of reference samples,when one pattern enters, the first phase of the inspection system to verify it.If it does not meet our requirements,we can judge it is false;if it have adopted the first phase,we use the second phase,Zernike moment to examine the characteristics.It has been proved that the identification of 2-pass system can save a lot of time for identification with out descending the correct identification rate.
引文
[1]L Yang,etc.Application of hidden markov models for signature identification [J].Pattern recognition,2005,28(2),pp.161-170
[2]J Dolfing,etal.On-line signature identification with hidden markov models [A].ICPR[C],2003,2 pp.1309-1312
[3]G Rigoll,A Kosmala,A systematic comparison between on-line and off-line methods for signature identification with hidden makov models[A],Proceedings of ICPR[C],1998.2,pp.1755-1757
[4]Q Z Wu,etc,On-line signature identification based on logarithmic spectrum [J].Pattern recognition,2001,31(12),pp.1865-1871
[5]R.plamondon and G.lerette,Automatic signature identification and writer identification-The state of the art,Pattern recognition,Vol.22,2006,pp.107-131
[6]Yi Liao Shu-Cherng etal.A neural network model with bounded-weights for pattern classification[J],Computers and Operations Research,2004,31(9):1411-1426.61
[7]Hai Ian,Hai-Zhou lli.Chinese signature verification with moment Invariants[A].IEEE InternaitonalC onferenceo nM ana ndC ybenreticsS ystems,12000,4:2963-2966
[8]侯海燕,脱机中文签名鉴定技术的研究,山东大学硕士学位论文,2007.5
[9]B.Fang,etal.Off-line signature verifieation by the traekingof feature and stroke positions.Pattern Reeognition,2003(36):91-101P
[10]M.Ammar,Progress in identification of skillfully simulated handwritten signature images,Pattern recognition Intell.5(1991),pp.337-351
[11]李军,面向对象的脱机签名验证系统,吉林大学硕士研究生学位论文,1994.4
[12]M.Ammar,Progress in identification of skillfully simulated handwritten signature images,Pattern recognition Intell.5(1991),pp.337-351
[13]Robert Sabourin,Member,IEEE,Ginette Genest,and Francoise J.Preteux Offiine Signature Identification System by Local Granulometric Size Distributions IEEE Transactions on Pattern Analysis and Machine Intelligence,Vol.19,no,1997
[14]F,Nouboud,Handwritten signature identification:a global approach,Fundamentals in Handwriting Identification Springer,Berlin,Heidelberg,1994,pp.445-459
[15]柯晶,乔谊正,赵宏,离线中文系统,计算机系统应用,1999,(2):7-10
[16]候卫萍,基于频域分析法的离线手写签名纹理特征提取和验证,哈尔滨工程大学硕士学位论文,2005
[17]程析,侯义斌,基于模糊模式识别的离线签名鉴别技术,计算机工程与应用,2001,37(13):79谢兆龙,脱机中文签名自动系统,硕士学位论文,济南:山东工业大学,1997.4-81
[18]朱勇,谭铁牛,王蕴红,基于笔记的身份鉴别,自动化学报,27(2):229-234,2001
[19]沈清、汤霖,模式识别导论,国防科技大学出版社,1991.5
[20]殷勤业,模式识别与神经网络,北京:机械工业出版社,1992.3
[21]胡守仁,余少波,戴葵,神经网络导论,国防科技大学出版社,1993.10
[22]C.Liu,R.Dai,Y.Liu,Extracting individual features from moments for Chinese writer identification,Proceedings of the Third IEEE International Conference on Document Analysis and Identification,Montreal,Canada,August 1995,pp.438-441
[23]R.Sabourin,M.Cheriet,G.Genest,An extended-shadow-code based approach for off-line signature identification,Proceedings of the Second ICDAR 93:International Conference on Document Analysis and Identification,Tsukuba Science City,Japan,1993,pp.1-6
[24]程析,侯义斌.基于模糊模式识别的离线签名鉴别技术.计算机工程与应用.2001,37(13):79-81
[25]左文明.脱机手写中文签名鉴别的研究.华南理工大学博士学位论文.2004:11-14
[26]G.Dimauro,S.Impedovo,A multi-expert signature identification system for bank-check processing,Int.J.Pattern recognition Artif.Intell.11(5)(1997),pp.827-844
[27]M.J.Revillet,Signature identification on postal cheques,in Proceedings of ICDAR,pp.763-773
[28]朱浩悦.脱机中文签名鉴别系统关键技术研究.西北大学硕士学位论文.2006:7-15
[29]王广松.脱机汉字签名鉴别研究.[研究生学位论文],福建:华侨大学,2004
[30]Khotanzad,A,Hong,Y.H.Invariant image Identification by Zernike moments Pattern Analysis and Machine Intelligence,IEEE Transactions onVolume 12,Issue 5,May 1990 Page(s),pp.489-497
[31]Hu M K,Visual Pattern recognition by Moment Invariants,IRE Trans Information Theory,1962,8,pp.179-187
[32]Wong Y R.Scene Matching with Invariant Moments Computer Graphics and Image Processing,1978,8,pp.16-24
[33]刘杰平,徐英林.一种简易的图像去噪方法.华南理工大学学报,2000,28(2):60-63
[34]方敏,徐俊艳,王建平等.一种新的文本图像二值化方法.合肥工业大学学报,2001,24(2):166-169
[35]赵宏,王丽敏,王工艺.汽车牌照自动鉴定中二值化方法的研究.应用科技,2004,31(3):15-19
[36]P.C.Chuang,Machine identification of handwritten signature image,In Proceedings of International Conference on Crime Countermeasure,pp.105-109,1977
[37]R.Sabourin and G.Genest,An extended-shadow-code based approach for offline signature identification ICDAR,1993,pp1-5
[38]Lianwen Jin,Gang WeilHandwritten Chinese Character Identification with Directional Decomposition Cellular Features[J]Journal of Circuit,System and Computer,1998,8(4),pp.517-524
[39]S W Lee,J S Park.Nonlinear Shape Normalization Methods for the Identification of Large Set Handwritten Character[J].Pattern recognition,1994,27(7),pp.895-902
[40]Sukhan Lee,Jack C.Pan Offline Tracing and Representation of Signatures Tansactions on Systems.Man.and Cybemetics Vol 22,No 4.July/August 1992,pp.755
[41]万亮,结合先验模型、无简单伪造训练样本的签名,硕士学位论文,西安:西北工业大学,2003.3
[42]Sabourin R,Drouhard J P,1992,Offiine signature identification using directional PDF and neural networks,Proceedings 11th international conference on pattern recognition,2:321-325
[43]Papamarkos.N.,Balatzakis.H.,Off-line Sigature Identification Using Multiple Nerual Network Classification Structures,International Conference on Digital Signal Processing,DSP,1997[C].IEEE,Piscataway,NJ,USA.p727-730
[44]A.El-Yacoubi,E.J.R.Justino et al(2000)Off-line signature identification using HMMs and cross-validation.2000,IEEE 859-868
[45]Khotanzad A,Hong Y H.Invariant Image Identification by Zernike Moment [J].IEEE Trans.PAMI,1990,12(5):489-497
[46]Hsu Y N,Arsenault H H,April G.Rotational Invariant Digital Pattern Using Circular Harmonic Expansion[J].Appl Opt,1982,(21):4 012-4 015
[47]M.Ammar,Progress in identification of skillfully simulated handwritten signature images,Pattern recognition Intell.5(1991),pp.337-351
[48]G Rigoll,A Kosmala,A systematic comparison between on-line and off-line methods for signature identification with hidden makov models[A],Proceedings of ICPR[C],1998.2,pp.1755-1757
[49]XiufenYe,WeiPingHouand WeixingFeng.Off-line Signature Verifieation With Infleetions Feature.Proeeedings of the 2005 ICIMA,China,2005:787-792P
[50]Reeves A P,Prokop R J,Andrews S E,et al.Three Dimensional Shape Analysis Using Moments and Fourier Descriptors[J].IEEE Trans.Pattern Anal.Machine Intell,1988,10(6):937-943
[51]牟小建,基于Zernike矩的脱机手写签名研究,[硕士学位论文],山东:山东大学,2006
[52]Hu M K.Visual Pattern recognition by Moment Invariant[J].IEEE Trans.Information Theory,1962,(8):179-187
[53]Persoon E,Fu K S.Shape Discrimination Using Fourier Descriptors[J].IEEE Trans.Syst Man Cybern,1997,(7):388-397
[54]谢兆龙,脱机中文签名自动系统,硕士学位论文,济南:山东工业大学,1997.4
[2]J Dolfing,etal.On-line signature identification with hidden markov models [A].ICPR[C],2003,2 pp.1309-1312
[3]G Rigoll,A Kosmala,A systematic comparison between on-line and off-line methods for signature identification with hidden makov models[A],Proceedings of ICPR[C],1998.2,pp.1755-1757
[4]Q Z Wu,etc,On-line signature identification based on logarithmic spectrum [J].Pattern recognition,2001,31(12),pp.1865-1871
[5]R.plamondon and G.lerette,Automatic signature identification and writer identification-The state of the art,Pattern recognition,Vol.22,2006,pp.107-131
[6]Yi Liao Shu-Cherng etal.A neural network model with bounded-weights for pattern classification[J],Computers and Operations Research,2004,31(9):1411-1426.61
[7]Hai Ian,Hai-Zhou lli.Chinese signature verification with moment Invariants[A].IEEE InternaitonalC onferenceo nM ana ndC ybenreticsS ystems,12000,4:2963-2966
[8]侯海燕,脱机中文签名鉴定技术的研究,山东大学硕士学位论文,2007.5
[9]B.Fang,etal.Off-line signature verifieation by the traekingof feature and stroke positions.Pattern Reeognition,2003(36):91-101P
[10]M.Ammar,Progress in identification of skillfully simulated handwritten signature images,Pattern recognition Intell.5(1991),pp.337-351
[11]李军,面向对象的脱机签名验证系统,吉林大学硕士研究生学位论文,1994.4
[12]M.Ammar,Progress in identification of skillfully simulated handwritten signature images,Pattern recognition Intell.5(1991),pp.337-351
[13]Robert Sabourin,Member,IEEE,Ginette Genest,and Francoise J.Preteux Offiine Signature Identification System by Local Granulometric Size Distributions IEEE Transactions on Pattern Analysis and Machine Intelligence,Vol.19,no,1997
[14]F,Nouboud,Handwritten signature identification:a global approach,Fundamentals in Handwriting Identification Springer,Berlin,Heidelberg,1994,pp.445-459
[15]柯晶,乔谊正,赵宏,离线中文系统,计算机系统应用,1999,(2):7-10
[16]候卫萍,基于频域分析法的离线手写签名纹理特征提取和验证,哈尔滨工程大学硕士学位论文,2005
[17]程析,侯义斌,基于模糊模式识别的离线签名鉴别技术,计算机工程与应用,2001,37(13):79谢兆龙,脱机中文签名自动系统,硕士学位论文,济南:山东工业大学,1997.4-81
[18]朱勇,谭铁牛,王蕴红,基于笔记的身份鉴别,自动化学报,27(2):229-234,2001
[19]沈清、汤霖,模式识别导论,国防科技大学出版社,1991.5
[20]殷勤业,模式识别与神经网络,北京:机械工业出版社,1992.3
[21]胡守仁,余少波,戴葵,神经网络导论,国防科技大学出版社,1993.10
[22]C.Liu,R.Dai,Y.Liu,Extracting individual features from moments for Chinese writer identification,Proceedings of the Third IEEE International Conference on Document Analysis and Identification,Montreal,Canada,August 1995,pp.438-441
[23]R.Sabourin,M.Cheriet,G.Genest,An extended-shadow-code based approach for off-line signature identification,Proceedings of the Second ICDAR 93:International Conference on Document Analysis and Identification,Tsukuba Science City,Japan,1993,pp.1-6
[24]程析,侯义斌.基于模糊模式识别的离线签名鉴别技术.计算机工程与应用.2001,37(13):79-81
[25]左文明.脱机手写中文签名鉴别的研究.华南理工大学博士学位论文.2004:11-14
[26]G.Dimauro,S.Impedovo,A multi-expert signature identification system for bank-check processing,Int.J.Pattern recognition Artif.Intell.11(5)(1997),pp.827-844
[27]M.J.Revillet,Signature identification on postal cheques,in Proceedings of ICDAR,pp.763-773
[28]朱浩悦.脱机中文签名鉴别系统关键技术研究.西北大学硕士学位论文.2006:7-15
[29]王广松.脱机汉字签名鉴别研究.[研究生学位论文],福建:华侨大学,2004
[30]Khotanzad,A,Hong,Y.H.Invariant image Identification by Zernike moments Pattern Analysis and Machine Intelligence,IEEE Transactions onVolume 12,Issue 5,May 1990 Page(s),pp.489-497
[31]Hu M K,Visual Pattern recognition by Moment Invariants,IRE Trans Information Theory,1962,8,pp.179-187
[32]Wong Y R.Scene Matching with Invariant Moments Computer Graphics and Image Processing,1978,8,pp.16-24
[33]刘杰平,徐英林.一种简易的图像去噪方法.华南理工大学学报,2000,28(2):60-63
[34]方敏,徐俊艳,王建平等.一种新的文本图像二值化方法.合肥工业大学学报,2001,24(2):166-169
[35]赵宏,王丽敏,王工艺.汽车牌照自动鉴定中二值化方法的研究.应用科技,2004,31(3):15-19
[36]P.C.Chuang,Machine identification of handwritten signature image,In Proceedings of International Conference on Crime Countermeasure,pp.105-109,1977
[37]R.Sabourin and G.Genest,An extended-shadow-code based approach for offline signature identification ICDAR,1993,pp1-5
[38]Lianwen Jin,Gang WeilHandwritten Chinese Character Identification with Directional Decomposition Cellular Features[J]Journal of Circuit,System and Computer,1998,8(4),pp.517-524
[39]S W Lee,J S Park.Nonlinear Shape Normalization Methods for the Identification of Large Set Handwritten Character[J].Pattern recognition,1994,27(7),pp.895-902
[40]Sukhan Lee,Jack C.Pan Offline Tracing and Representation of Signatures Tansactions on Systems.Man.and Cybemetics Vol 22,No 4.July/August 1992,pp.755
[41]万亮,结合先验模型、无简单伪造训练样本的签名,硕士学位论文,西安:西北工业大学,2003.3
[42]Sabourin R,Drouhard J P,1992,Offiine signature identification using directional PDF and neural networks,Proceedings 11th international conference on pattern recognition,2:321-325
[43]Papamarkos.N.,Balatzakis.H.,Off-line Sigature Identification Using Multiple Nerual Network Classification Structures,International Conference on Digital Signal Processing,DSP,1997[C].IEEE,Piscataway,NJ,USA.p727-730
[44]A.El-Yacoubi,E.J.R.Justino et al(2000)Off-line signature identification using HMMs and cross-validation.2000,IEEE 859-868
[45]Khotanzad A,Hong Y H.Invariant Image Identification by Zernike Moment [J].IEEE Trans.PAMI,1990,12(5):489-497
[46]Hsu Y N,Arsenault H H,April G.Rotational Invariant Digital Pattern Using Circular Harmonic Expansion[J].Appl Opt,1982,(21):4 012-4 015
[47]M.Ammar,Progress in identification of skillfully simulated handwritten signature images,Pattern recognition Intell.5(1991),pp.337-351
[48]G Rigoll,A Kosmala,A systematic comparison between on-line and off-line methods for signature identification with hidden makov models[A],Proceedings of ICPR[C],1998.2,pp.1755-1757
[49]XiufenYe,WeiPingHouand WeixingFeng.Off-line Signature Verifieation With Infleetions Feature.Proeeedings of the 2005 ICIMA,China,2005:787-792P
[50]Reeves A P,Prokop R J,Andrews S E,et al.Three Dimensional Shape Analysis Using Moments and Fourier Descriptors[J].IEEE Trans.Pattern Anal.Machine Intell,1988,10(6):937-943
[51]牟小建,基于Zernike矩的脱机手写签名研究,[硕士学位论文],山东:山东大学,2006
[52]Hu M K.Visual Pattern recognition by Moment Invariant[J].IEEE Trans.Information Theory,1962,(8):179-187
[53]Persoon E,Fu K S.Shape Discrimination Using Fourier Descriptors[J].IEEE Trans.Syst Man Cybern,1997,(7):388-397
[54]谢兆龙,脱机中文签名自动系统,硕士学位论文,济南:山东工业大学,1997.4
© 2004-2018 中国地质图书馆版权所有 京ICP备05064691号 京公网安备11010802017129号 地址：北京市海淀区学院路29号 邮编：100083 电话：办公室：(+86 10)66554848；文献借阅、咨询服务、科技查新：66554700 |