自动指纹识别中关键算法的研究
|
摘要
近年来，随着社会和经济的发展，人们对身份鉴别的准确性、安全性和实用性提出了更高的要求。生物识别技术已生物特征为基础，以信息处理技术为手段，将生物技术和信息技术有机的机构在一起。在众多生物识别技术中，指纹识别以方便易用、高准确率、易采集和低成本等诸多优势备受关注，已经成为身份认证的最有效手段，在犯罪识别、信息安全、电子商务等领域得到广泛的应用。
自动指纹识别系统是集光电技术、图像处理、计算机及网络、数据库技术、模式识别技术等于一体的综合性系统。针对自动指纹识别系统的运行过程，在总结该领域现有研究成果的基础上，本文对自动指纹识别系统中的指纹图像预处理、指纹特征提取和分类以及指纹匹配的关键算法进行了大量的研究，主要研究和结果如下：
(1)利用Gaussian滤波器对指纹图像多次滤波计算指纹图像方向图，减小了噪声的影响；根据得到的方向信息对原有图像进行对应分块旋转并沿Y轴垂直投影得到图像的统计频率，算法简便；采用mean-shift算法进行指纹增强，取整幅图像有效频率的均值作为该图像的频率，减少了计算量。并在此基础上利用一定的判别规则对指纹图像进行可信度判别，进一步改进了增强后图像的质量。该算法简单有效，易于实现。
(2)提取指纹的细节特征点并保存真实特征点的属性。采用支持向量机理论的两类指纹分类方法。SVM方法基于严密的数学理论，遵循SRM原则寻找最优超平面，表现出了很好的泛化能力，充分发挥了SVM理论解决二类分类问题的优势，训练样本具有较好的推广能力。
(3)利用遗传算法进行细节点匹配。用改进的基于遗传算法的点匹配算法求出使对应点数目最多、匹配误差最小的细节点对应关系，根据得到的对应点数目和匹配误差大小推理出匹配分值。
最后总结了全文，分析了目前研究工作中需要进一步完善的地方，指出今后工作的研究方向。
Accurate, secure and practical personal identification methods are highly required with the development of the social and economy. Biometrics based on the physiological or behavior traits identification provides a convenient and reliable scheme. Among the numerous biometrics, more attentions have been paid to the fingerprint identification technology due to its convenience, high accuracy and low cost. The technique of fingerprint identification has become one of the widest used biometric identification techniques. It has been widely used in electronic commerce,criminal identification, information safety etc.
Automated Fingerprint Identification System is an integrated system which concentrate the photoelectric technology, image processing, computer and networking, database technology, pattern recognition technology. In the view of the operation of the APIS, based on the conclusion of the existing research results, I research the key algorithm in the fingerprint image pretreatment preprocessing, feature extraction, fingerprint classifier and feature matching algorithm of the fingerprint authentication system in this paper. Major research and the results are as follows:
(1)The orientation is calculated by repetitious Gaussian filtering on fingerprint images through which noises can be removed well. According to the founded orientations, we rotate the original images divided by corresponding blocks. It is convenient to get the statistical frequencies by projecting the rotated blocks along axis Y. Fingerprint enhancement is processed through Gabor filter. And the processing time is reduced by using the average value of effective frequencies as the frequency of the whole image. A discriminant based on reliability is also adopted to improved the binary images. The algorithm presented is simple and effective, and easier to realized.
(2) The existed minutiae extraction and classification algorithms. A tow-stage fingerprint classification algorithm based on the Support Vector Machine is presented. The SVM method is based on strict mathematical theory, find the optimal SRM hyperplane guided by the principle, displayed good generalization ability, the sufficiency uses SVM theory which the advantages take on the second category classification. The training sample have better promoting capacity.
(3) A minutiae matching method combining using genetic algorithm is proposed. Using a modified points matching method based on genetic algorithm to find the best minutiae corresponding relations which having the most matching pairs and minimum matching error. Then reason the matching score based on defined the relations between the number of matching pairs and the matching error.
Finally, according to summary of this thesis, the improvements need to be finished are analyzed, and the direction of future work is given.
自动指纹识别系统是集光电技术、图像处理、计算机及网络、数据库技术、模式识别技术等于一体的综合性系统。针对自动指纹识别系统的运行过程，在总结该领域现有研究成果的基础上，本文对自动指纹识别系统中的指纹图像预处理、指纹特征提取和分类以及指纹匹配的关键算法进行了大量的研究，主要研究和结果如下：
(1)利用Gaussian滤波器对指纹图像多次滤波计算指纹图像方向图，减小了噪声的影响；根据得到的方向信息对原有图像进行对应分块旋转并沿Y轴垂直投影得到图像的统计频率，算法简便；采用mean-shift算法进行指纹增强，取整幅图像有效频率的均值作为该图像的频率，减少了计算量。并在此基础上利用一定的判别规则对指纹图像进行可信度判别，进一步改进了增强后图像的质量。该算法简单有效，易于实现。
(2)提取指纹的细节特征点并保存真实特征点的属性。采用支持向量机理论的两类指纹分类方法。SVM方法基于严密的数学理论，遵循SRM原则寻找最优超平面，表现出了很好的泛化能力，充分发挥了SVM理论解决二类分类问题的优势，训练样本具有较好的推广能力。
(3)利用遗传算法进行细节点匹配。用改进的基于遗传算法的点匹配算法求出使对应点数目最多、匹配误差最小的细节点对应关系，根据得到的对应点数目和匹配误差大小推理出匹配分值。
最后总结了全文，分析了目前研究工作中需要进一步完善的地方，指出今后工作的研究方向。
Accurate, secure and practical personal identification methods are highly required with the development of the social and economy. Biometrics based on the physiological or behavior traits identification provides a convenient and reliable scheme. Among the numerous biometrics, more attentions have been paid to the fingerprint identification technology due to its convenience, high accuracy and low cost. The technique of fingerprint identification has become one of the widest used biometric identification techniques. It has been widely used in electronic commerce,criminal identification, information safety etc.
Automated Fingerprint Identification System is an integrated system which concentrate the photoelectric technology, image processing, computer and networking, database technology, pattern recognition technology. In the view of the operation of the APIS, based on the conclusion of the existing research results, I research the key algorithm in the fingerprint image pretreatment preprocessing, feature extraction, fingerprint classifier and feature matching algorithm of the fingerprint authentication system in this paper. Major research and the results are as follows:
(1)The orientation is calculated by repetitious Gaussian filtering on fingerprint images through which noises can be removed well. According to the founded orientations, we rotate the original images divided by corresponding blocks. It is convenient to get the statistical frequencies by projecting the rotated blocks along axis Y. Fingerprint enhancement is processed through Gabor filter. And the processing time is reduced by using the average value of effective frequencies as the frequency of the whole image. A discriminant based on reliability is also adopted to improved the binary images. The algorithm presented is simple and effective, and easier to realized.
(2) The existed minutiae extraction and classification algorithms. A tow-stage fingerprint classification algorithm based on the Support Vector Machine is presented. The SVM method is based on strict mathematical theory, find the optimal SRM hyperplane guided by the principle, displayed good generalization ability, the sufficiency uses SVM theory which the advantages take on the second category classification. The training sample have better promoting capacity.
(3) A minutiae matching method combining using genetic algorithm is proposed. Using a modified points matching method based on genetic algorithm to find the best minutiae corresponding relations which having the most matching pairs and minimum matching error. Then reason the matching score based on defined the relations between the number of matching pairs and the matching error.
Finally, according to summary of this thesis, the improvements need to be finished are analyzed, and the direction of future work is given.
引文
[1] 孙东梅，裘正定。生物特征识别技术综述。电子学报，2001，29(12)：1744-1748
[2] 张敏贵等。生物特征识别及研究现状。生物物理学报，2002，18(2)；156-162。
[3] R Clarke. Human identification in information systems; Management challenges and public policy issues. Info Techno People, 1994,7(4):3-37.
[4] Anil. K. Jain, An Identity-Authentication System Using Fingerprint, Proceeding of The IEEE, Voi. 85, No. 9, September 1997
[5] 尹义龙．“自动指纹识别技术的发展与应用”．南京大学学报(自然科学版)．2002，38(1)：29-35
[6] 冯星奎，李林艳，颜祖泉．一种新的指纹图像细化算法[J]．中国图像图形学报。1999，14(10)：835-838．
[7] Jeong-woo, Lee, etc. A 600 dpi capacitive fingerprint intechniuqe, Solid-StateCircuits , IEEE, 1999,344:469-475
[8] X. Jiang and W. Yau. Fingerprint Minutiae Matching Based on the Local And Global Structures, ICPR2000, Spain, 2000.
[9] Xiao Qing-han, Hazam Fingerprint Images Preprocessing: A combined statistical Recognition 1991,24(10):985-989. Hung 1993, Pattern
[10] A. Roddy and J. Stosz. Fingerprint features statistical analysis and system performance estimates [J]. Proceedings of IEEE, 1997, 85(9):1390-1421
[11] Shesha Shah and P Sastry. Fingerprint classification using a feedback-based line detector[J]. IEEE Trans. Transactions on system. 2004 34(1):85-94.
[12] 祁亨年．支持向量机及其应用研究综述[J]．计算机工程．2004，30(10)：6-9．
[13] 罗希平，田捷．自动指纹识别中的图像增强和细节匹配算法[J]．软件学报，2002，1 3(05)：946-953．
[14] 程建刚：田捷等．基于非线性扩散滤波的指纹增强算法[J]．自动化学报，2004，30(6)：854-862．
[15] 蔡俊，任德官．基于BP神经网络的指纹模板分类器分类算法[J]。微电子学与计算机．2002，(9)：1-3．
[16] Cheng YZ. Mean shift, mode seeking, and clustering. IEEE Trans. on Pattern Analysis and Machine Intelligence, 1995,17(8): 790-799. (1995)
[17] Comaniciu D, MeeT P. Mean shift: A robust approach toward feature spaceanal ysis. IEEE Trans. on Pattern Analysis and Machine Intelligence, 2002,24(5):603-619.
[18] Comaniciu D, Ramesh V, Meet P. Real-time tracking of non-rigid objects using mean shift. Computer Vision and Pattern Recognition, 2000.142～149
[19] Collins RT. Mean-Shift blob tracking through scale space. In: Proc. of the Conf. on Computer Vision and Pattern Recognition (CVPR). 2003.18-20.
[20] Comaniciu D. An algorithm for data-driven bandwidth selection. IEEE Trans. on Pattern Analysis and Machine Intelligence, 2003, 25(2):281-288.
[21] Lee H C, Gaensslen R E. Advances in Fingerprint Technology [M], New York: Elsevier, 1991:22-24.
[22] Rappelli and A Lumini. Fingerprint classification by directional image partitioning[J]. IEEE Trans. Pattern Anal. and Machine Inteli. 1999,21(5):256-261.
[23] Sen Wang, Wei Wei, Zhang Yang, et al. Fingerprint classification by directional fields[C]. Pro. of the4 IEEE International Conference on Multimodal Interfaces. 2002.
[24] T K Moon and W C Stirling. Mathematical methods and algorithms for signal processing [M]. Upper Sadle River: Prentice Hall, 1999.
[25] 王崇文，李见为，陈为民．基于HMM和SVM的指纹分类方法[J]．电子与信息学报．2003，(25)：1488-1493．
[26] Anil K. Jain, Salil Prabhakar, Lin Hong. A multichannel approach to fingerprint classification[J].IEEE Transaction on Pattern Analysis and Machine Intelligence. 1999, 21(4):34-359.
[27] N. Ratha, S. Chen, and A.K. Jain. Adaptive flow orientation based feature extraction in fingerprint images[J]. Pattern Recognition.1995，28(11)：165—1672．
[28] 杨小冬，宁新宝，詹小四等．基于纹线跟踪的指纹分类方法[J]．计算机工 程．2005，3l(7)：170-173．
[29] X. Jiang, WY. Yau andW. Ser. Minutiae extraction by adaptive tracing the gray level ridge of the fingerprint image[C]. IEEE 6~(th) International Conference on Image Processing, Japan. 1999: 852-856.
[30] 尹义龙，宁新宝，张晓梅．改进的指纹细节特征提取算法[[J]．中国图象图形学报．2002，17(12)：1302-1306．
[31] 崔凤奎，王晓强，张丰等．二值图像细化算法的比较与改进[J]．洛阳工学院学报．1997，18(4)：48-52．
[32] 卢的，陶刚，李吉桂．一种新的指纹图像细化算法[J]．现代计算机．2002．N0．151：42-44．
[33] 王崇文，王伟，王廷才等．指纹图像后处理[J]．计算机工程与设计．2002，23(9)：35-38．
[34] 刘怀泉．自动指纹识别系统研究[D]．硕士学位论文，山东大学，2005
[35] 邵志勇，温春友，田莹．指纹图像二值化方法的比较研究[J]．鞍山科技大学学报．2004，27(3)：186-189．
[36] 苑玮琦，夏义勇．方向滤波指纹图像二值化方法[J]．仪器仪表学报．200324(4)增刊：469-471．
[37] 李建华，马小妹，郭成安，基于方向图的动态阈值指纹图像二值化方法[J]．大连理工大学学报．2002，42(5)：626-628．
[38] 罗希平，田捷．自动指纹识别中的图像增强和细节匹配算法[J]．软件学报．2002，13(5)：946-956．
[39] 柯渝．指纹识别系统中的预处理与后处理研究[M]。硕士学位论文，厦门大学计算机系，2004．6．
[40] M M Chong. Geometric framework for fingerprint image classification [J].Pattern Recognition. 1997, 30(9): 147—1488.
[2] 张敏贵等。生物特征识别及研究现状。生物物理学报，2002，18(2)；156-162。
[3] R Clarke. Human identification in information systems; Management challenges and public policy issues. Info Techno People, 1994,7(4):3-37.
[4] Anil. K. Jain, An Identity-Authentication System Using Fingerprint, Proceeding of The IEEE, Voi. 85, No. 9, September 1997
[5] 尹义龙．“自动指纹识别技术的发展与应用”．南京大学学报(自然科学版)．2002，38(1)：29-35
[6] 冯星奎，李林艳，颜祖泉．一种新的指纹图像细化算法[J]．中国图像图形学报。1999，14(10)：835-838．
[7] Jeong-woo, Lee, etc. A 600 dpi capacitive fingerprint intechniuqe, Solid-StateCircuits , IEEE, 1999,344:469-475
[8] X. Jiang and W. Yau. Fingerprint Minutiae Matching Based on the Local And Global Structures, ICPR2000, Spain, 2000.
[9] Xiao Qing-han, Hazam Fingerprint Images Preprocessing: A combined statistical Recognition 1991,24(10):985-989. Hung 1993, Pattern
[10] A. Roddy and J. Stosz. Fingerprint features statistical analysis and system performance estimates [J]. Proceedings of IEEE, 1997, 85(9):1390-1421
[11] Shesha Shah and P Sastry. Fingerprint classification using a feedback-based line detector[J]. IEEE Trans. Transactions on system. 2004 34(1):85-94.
[12] 祁亨年．支持向量机及其应用研究综述[J]．计算机工程．2004，30(10)：6-9．
[13] 罗希平，田捷．自动指纹识别中的图像增强和细节匹配算法[J]．软件学报，2002，1 3(05)：946-953．
[14] 程建刚：田捷等．基于非线性扩散滤波的指纹增强算法[J]．自动化学报，2004，30(6)：854-862．
[15] 蔡俊，任德官．基于BP神经网络的指纹模板分类器分类算法[J]。微电子学与计算机．2002，(9)：1-3．
[16] Cheng YZ. Mean shift, mode seeking, and clustering. IEEE Trans. on Pattern Analysis and Machine Intelligence, 1995,17(8): 790-799. (1995)
[17] Comaniciu D, MeeT P. Mean shift: A robust approach toward feature spaceanal ysis. IEEE Trans. on Pattern Analysis and Machine Intelligence, 2002,24(5):603-619.
[18] Comaniciu D, Ramesh V, Meet P. Real-time tracking of non-rigid objects using mean shift. Computer Vision and Pattern Recognition, 2000.142～149
[19] Collins RT. Mean-Shift blob tracking through scale space. In: Proc. of the Conf. on Computer Vision and Pattern Recognition (CVPR). 2003.18-20.
[20] Comaniciu D. An algorithm for data-driven bandwidth selection. IEEE Trans. on Pattern Analysis and Machine Intelligence, 2003, 25(2):281-288.
[21] Lee H C, Gaensslen R E. Advances in Fingerprint Technology [M], New York: Elsevier, 1991:22-24.
[22] Rappelli and A Lumini. Fingerprint classification by directional image partitioning[J]. IEEE Trans. Pattern Anal. and Machine Inteli. 1999,21(5):256-261.
[23] Sen Wang, Wei Wei, Zhang Yang, et al. Fingerprint classification by directional fields[C]. Pro. of the4 IEEE International Conference on Multimodal Interfaces. 2002.
[24] T K Moon and W C Stirling. Mathematical methods and algorithms for signal processing [M]. Upper Sadle River: Prentice Hall, 1999.
[25] 王崇文，李见为，陈为民．基于HMM和SVM的指纹分类方法[J]．电子与信息学报．2003，(25)：1488-1493．
[26] Anil K. Jain, Salil Prabhakar, Lin Hong. A multichannel approach to fingerprint classification[J].IEEE Transaction on Pattern Analysis and Machine Intelligence. 1999, 21(4):34-359.
[27] N. Ratha, S. Chen, and A.K. Jain. Adaptive flow orientation based feature extraction in fingerprint images[J]. Pattern Recognition.1995，28(11)：165—1672．
[28] 杨小冬，宁新宝，詹小四等．基于纹线跟踪的指纹分类方法[J]．计算机工 程．2005，3l(7)：170-173．
[29] X. Jiang, WY. Yau andW. Ser. Minutiae extraction by adaptive tracing the gray level ridge of the fingerprint image[C]. IEEE 6~(th) International Conference on Image Processing, Japan. 1999: 852-856.
[30] 尹义龙，宁新宝，张晓梅．改进的指纹细节特征提取算法[[J]．中国图象图形学报．2002，17(12)：1302-1306．
[31] 崔凤奎，王晓强，张丰等．二值图像细化算法的比较与改进[J]．洛阳工学院学报．1997，18(4)：48-52．
[32] 卢的，陶刚，李吉桂．一种新的指纹图像细化算法[J]．现代计算机．2002．N0．151：42-44．
[33] 王崇文，王伟，王廷才等．指纹图像后处理[J]．计算机工程与设计．2002，23(9)：35-38．
[34] 刘怀泉．自动指纹识别系统研究[D]．硕士学位论文，山东大学，2005
[35] 邵志勇，温春友，田莹．指纹图像二值化方法的比较研究[J]．鞍山科技大学学报．2004，27(3)：186-189．
[36] 苑玮琦，夏义勇．方向滤波指纹图像二值化方法[J]．仪器仪表学报．200324(4)增刊：469-471．
[37] 李建华，马小妹，郭成安，基于方向图的动态阈值指纹图像二值化方法[J]．大连理工大学学报．2002，42(5)：626-628．
[38] 罗希平，田捷．自动指纹识别中的图像增强和细节匹配算法[J]．软件学报．2002，13(5)：946-956．
[39] 柯渝．指纹识别系统中的预处理与后处理研究[M]。硕士学位论文，厦门大学计算机系，2004．6．
[40] M M Chong. Geometric framework for fingerprint image classification [J].Pattern Recognition. 1997, 30(9): 147—1488.
© 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 |