基于特征点局部结构相似性的指纹快速匹配方法的研究
|
摘要
与其他生物特征识别技术相比,指纹特征具有唯一性和稳定性等优点,指纹识别技术是目前应用最广泛的生物特征识别技术,具有悠久的历史。特别是随着计算机信息处理技术的飞速发展,为自动指纹识别的应用开拓了更为广阔的市场,指纹识别技术与相关产品越来越多地应用于嵌入式设备中,同时也对指纹识别算法提出了新的挑战和要求,可靠和快速的嵌入式指纹识别系统有了更迫切的需求。本文对自动指纹识别过程中的一些关键算法进行了研究和改进,主要包括指纹快速增强和快速匹配算法的研究和改进,从而改善指纹识别系统的整体性能。?
指纹图像增强是指纹预处理中非常关键的一步,其增强效果的好坏直接影响到指纹识别的后续处理过程。本文在原有传统二维Gabor滤波指纹增强方法的基础上,通过把二维Gabor滤波器分解为一维Gauss滤波器的组合,对现有的指纹图像增强算法进行改进,二维Gabor滤波的一维分解实现能极大地降低计算复杂度,减少运算次数。实验结果表明,改进后的指纹增强算法运行时间大大减少,也能保持指纹图像的增强效果。?
指纹图像匹配是指纹识别的最后一步,也一直是指纹识别系统最为核心的研究内容之一。基于点模式的指纹匹配由于存储容量小,安全性高等优点,一直是研究的热点,但很多指纹匹配方法存在计算复杂度大,匹配速度慢,匹配精度低等问题,可靠和快速的指纹匹配在自动指纹识别系统中仍然是一个挑战。本文提出了一种改进的基于细节点的指纹快速匹配算法,构建以细节点为中心的环形区域作为特征点的局部结构来确保匹配的速度并对非线性形变有较好的鲁棒性。因为每个环形区域是指纹的一个很小的区域,所以通过这个环形区域,可以更快的找到匹配的细节点。一般而言,特征点局部结构相似性的可靠性更高,所以与全局匹配方法相比,本文方法对非线性形变有更好的鲁棒性。此外,构建的环形区域比局部匹配方法构建的局部近邻区域更大,所以构建的环形区域比局部匹配方法构建的局部邻域包含更多的细节点,匹配结果更可靠。实验结果表明,该方法在处理速度和准确性上有更好的性能。?
在指纹识别系统中应用这两种改进方法,采用FVC2004的四个数据库进行实验的结果表明,本文实现的指纹快速增强和快速匹配方法能极大地提高指纹识别的速度。?
Compared to other biometric identification technology, fingerprint recognition has the advantages of uniqueness and stability, Fingerprint recognition as the most widely used Biometric Authentication has a long history, Especially, with the rapid development of information processing technology, fingerprint identification obtains a wider market and are increasingly used in embedded devices. At the same time, Fingerprint recognition algorithm confronts the new challenges and demands. Reliable and fast fingerprint identification has more pressing needs. This paper is about the study and improvement of some key algorithms of automatic fingerprint recognition, including the fast enhancement and matching of fingerprint image, thus, improving the overall performance of fingerprint identification system.
Fingerprint image enhancement is a very step of fingerprint preprocessing, the enhanced effect will have a direct impact on the follow-up process of the fingerprint identification. Based on the traditional two-dimensional Gabor filter enhancement, the paper improved the existing fingerprint image enhancement algorithm by dividing two-dimensional Gabor filter into the combination of one-dimensional Gauss filter. The implement of one-dimension decomposition of two-dimensional Gabor filter can greatly reduce the complexity and the number of operations. Experimental result shows that the improved fingerprint enhancement algorithm can reduce the running time greatly and also maintain the effect of fingerprint enhancement.
Fingerprint image matching algorithm as the final step in fingerprint recognition is also the one of the most core research content. Because of the small storage template and high security, Fingerprint matching based on point pattern has been a hot research. However, many fingerprint matching methods have the disadvantages of large computational complexity, slow matching speed and low matching precision, Reliable and fast fingerprint matching in the automated fingerprint identification system remains a challenge. This paper presents an improved minutiae-based fingerprint fast matching algorithm which can speed up the matching speed and ensure the robustness to non-linear deformation by building the minutiae-centered annular region as the local structure of feature points. We can find the matching minutiaes more quickly through the annular region because it is a small region of fingerprint. In general, the local structure similarity of feature points has higher reliability. Therefore, compared with the global match, the improved method will have better robustness to the nonlinear deformation. In addition, the annular area is larger than the local neighbors area of the local matching, so the annular area contains more minutiaes than the local neighborhood of the local matching, thus, the matching results will be more reliable. Experimental result shows that the improved matching method has better performance in processing rate and accuracy.
We apply these two improved methods in the fingerprint identification system, experimental result using the four databases of FVC2004 shows that the improved fingerprint rapid enhancement and fast matching methods can greatly improve the speed of fingerprint identification.
指纹图像增强是指纹预处理中非常关键的一步,其增强效果的好坏直接影响到指纹识别的后续处理过程。本文在原有传统二维Gabor滤波指纹增强方法的基础上,通过把二维Gabor滤波器分解为一维Gauss滤波器的组合,对现有的指纹图像增强算法进行改进,二维Gabor滤波的一维分解实现能极大地降低计算复杂度,减少运算次数。实验结果表明,改进后的指纹增强算法运行时间大大减少,也能保持指纹图像的增强效果。?
指纹图像匹配是指纹识别的最后一步,也一直是指纹识别系统最为核心的研究内容之一。基于点模式的指纹匹配由于存储容量小,安全性高等优点,一直是研究的热点,但很多指纹匹配方法存在计算复杂度大,匹配速度慢,匹配精度低等问题,可靠和快速的指纹匹配在自动指纹识别系统中仍然是一个挑战。本文提出了一种改进的基于细节点的指纹快速匹配算法,构建以细节点为中心的环形区域作为特征点的局部结构来确保匹配的速度并对非线性形变有较好的鲁棒性。因为每个环形区域是指纹的一个很小的区域,所以通过这个环形区域,可以更快的找到匹配的细节点。一般而言,特征点局部结构相似性的可靠性更高,所以与全局匹配方法相比,本文方法对非线性形变有更好的鲁棒性。此外,构建的环形区域比局部匹配方法构建的局部近邻区域更大,所以构建的环形区域比局部匹配方法构建的局部邻域包含更多的细节点,匹配结果更可靠。实验结果表明,该方法在处理速度和准确性上有更好的性能。?
在指纹识别系统中应用这两种改进方法,采用FVC2004的四个数据库进行实验的结果表明,本文实现的指纹快速增强和快速匹配方法能极大地提高指纹识别的速度。?
Compared to other biometric identification technology, fingerprint recognition has the advantages of uniqueness and stability, Fingerprint recognition as the most widely used Biometric Authentication has a long history, Especially, with the rapid development of information processing technology, fingerprint identification obtains a wider market and are increasingly used in embedded devices. At the same time, Fingerprint recognition algorithm confronts the new challenges and demands. Reliable and fast fingerprint identification has more pressing needs. This paper is about the study and improvement of some key algorithms of automatic fingerprint recognition, including the fast enhancement and matching of fingerprint image, thus, improving the overall performance of fingerprint identification system.
Fingerprint image enhancement is a very step of fingerprint preprocessing, the enhanced effect will have a direct impact on the follow-up process of the fingerprint identification. Based on the traditional two-dimensional Gabor filter enhancement, the paper improved the existing fingerprint image enhancement algorithm by dividing two-dimensional Gabor filter into the combination of one-dimensional Gauss filter. The implement of one-dimension decomposition of two-dimensional Gabor filter can greatly reduce the complexity and the number of operations. Experimental result shows that the improved fingerprint enhancement algorithm can reduce the running time greatly and also maintain the effect of fingerprint enhancement.
Fingerprint image matching algorithm as the final step in fingerprint recognition is also the one of the most core research content. Because of the small storage template and high security, Fingerprint matching based on point pattern has been a hot research. However, many fingerprint matching methods have the disadvantages of large computational complexity, slow matching speed and low matching precision, Reliable and fast fingerprint matching in the automated fingerprint identification system remains a challenge. This paper presents an improved minutiae-based fingerprint fast matching algorithm which can speed up the matching speed and ensure the robustness to non-linear deformation by building the minutiae-centered annular region as the local structure of feature points. We can find the matching minutiaes more quickly through the annular region because it is a small region of fingerprint. In general, the local structure similarity of feature points has higher reliability. Therefore, compared with the global match, the improved method will have better robustness to the nonlinear deformation. In addition, the annular area is larger than the local neighbors area of the local matching, so the annular area contains more minutiaes than the local neighborhood of the local matching, thus, the matching results will be more reliable. Experimental result shows that the improved matching method has better performance in processing rate and accuracy.
We apply these two improved methods in the fingerprint identification system, experimental result using the four databases of FVC2004 shows that the improved fingerprint rapid enhancement and fast matching methods can greatly improve the speed of fingerprint identification.
引文
[1]陈桂友,孙同景.自动指纹识别系统中的关键算法研究及应用[D].济南:山东大学,2005年10月
[2]卢官明等.生物特征识别综述[J].南京邮电大学学报, 2007, 27(1):82~87
[3]彭章平.自动指纹识别系统研究[D].长沙:中南大学
[4]张新淼.基于改进的Gabor滤波器组指纹图像识别的研究和实现[D].天津:天津师范大学, 2009年3月
[5]付景广.指纹识别中若干关键算法的研究[D].北京:中国科学院软件研究所,2003
[6]汪龙峰.低质量指纹图像增强与形变指纹匹配的研究[D].长沙:中南大学, 2009
[7]张宇.指纹图像质量评价方法研究[D].济南:山东大学, 2009
[8]周晔华.指纹图像预处理算法研究[D].南京:南京理工大学,2008
[9]杨雪.指纹识别系统研究[D].哈尔滨:哈尔滨理工大学, 2009
[10]夏振华,石玉,于盛林.基于Gabor滤波器的指纹增强[J].工程图像学报, 2006(5):81~85
[11] B.G.Sherlock, D.M. Monro, K. Millard. Fingerprint enhancement by directional Fourier filtering[J]. IEE hoc.-Vis. Image Signal Process, 1994, 141(2):87~93
[12]傅景广,许刚,王裕国.基于二值图像的指纹细节点提取[J].计算机研究与发展,2004, 41(4):721-727
[13] Nalini IS. Ratha &Vinayaka D. Pandit. Robust Fingerprint Authentication Using Local Structural Similarity[J]. IEEE,2000:29-34
[14] M. M. Hadhoud, W. S. ElKilani, M. I. Samaan. An Adaptive Algorithm for Fingerprints Image Enhancement Using Gabor Filters[J]. IEEE , 2007:227-236
[15]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):855-862
[16] Lin Hong, Yifei Wan, and Anil Jain . Fingerprint Image Enhancement:Algorithm and Performance Evaluation. IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE, 1998, 20(8):777-789
[17]张建伟,陆亨立.指纹自动识别中图像分割方法的研究[J].微型电脑应用, 1999, 15(12) : 20-22
[18]黄贤武,苏鹏程,柏培权.基于方向滤波分割的指纹自动识别系统算法[J].中国图像图形报.2002 ,7 (8) :829.
[19]蒙丽彬,赵进创,傅文利.一种改进的基于G abor滤波的指纹增强算法[J].微计算机信息. 2006, 22(11-3):297-299
[20]叶四民,陈福祥.指纹图像预处理中的二值化技术[J].自动化与仪器仪表, 2001(2):30-39
[21]冯星奎,李林艳,颜祖泉.一种新的指纹图象细化算法[J].中国图像图形学报, 1999,4(10):835-837
[22]杨凡,赵顺东.一种有效的混合式指纹快速细化算法[J]. 2008, 25(10):3035-3041
[23]马宁.指纹图像的二值化与细化研究[D].南京:南京理工大学,2006
[24] Louisa Lam, Seong-Whan Lee. Thinning Methodologies-A Comprehensive Survey. IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE, 1992,14(9): 869-882
[25]王业琳,宁新宝,尹义龙.指纹图像细化算法的研究[J].南京大学学报. 2003, 39(4):468-475
[26] T.Y. Zhang & C.Y. Suen A fast Parallel thinning algorithm for thinning digital pattern. Communication of ACM.1984,27(3):236-239
[27]李小平,边肇祺,汪云九.二维Gabor滤波器的快速实现[J],自动化学报, 1989, 15(2):136-140
[28] Jan-Mark Geusebroek, Arnold W. M. Smeulders. Fast Anisotropic Gauss Filtering[J], IEEE TRANSACTIONS ON IMAGE PROCESSING,2003, 12(8):938-943
[29] Sanja Ranade, Azriel Rosenfeld. Point pattern matching by relaxation.Pattern Recognition[J],1980,12(4):269-275
[30] G.Stockman,S.Kopstein,S.Benett.Matching images to models for registration and object detection via clusterning[J] .IEEE Trans. On Pattern Analysis and Machine Intelligence,1982,4(3):229-241
[31] J.P.Pascual Starink, Eric Backer. Finding point correspondences using simulated annealing[J] .Pattern Recognition,1995,28(2):231-240
[32] A.K.Jain, Lin Hong, Sharath Pankanti, Ruud Bolle. An identity-anthentication system using fingerprints[J]. Proceedings of the IEEE,1997,85(9):1365-1388
[33]罗喜平,田捷.自动指纹识别总的图像增强和图像匹配[J].软件学报,2002,13(5):946-56
[34] Yuliang He, Jie Tian,Xiping Luo, Tanghui Zhang. Image enhancement and minutia matching in fingerprint verification[J].Pattern Recognition Letter,2003,24(9):1349-1360
[35] Xudong Jiang, Wei-Yun Yau. Fingerprint minutiae matching based on the local and global structures[J] .IEEE,2000:1038-1041
[36] C.I.Watson, P.J.Grother, D.P.Casasent. Distortion-tolerant fileter for elastic-distorted fingerprint matching[J].SPIE,2001
[37] Jain AK, Prabhakar S, Hong L, Pankanti S. Filterbank-based Fingerprint matching[J].IEEE Transaction On image Processing, 2000,9(5):846~859
[38] Anil Jain, Arun Ross, Salil Prabhakar. fingerprint matching using minutiae and texture features[J],in Proc. International Conference on Image Processing(ICIP), 2001:282-285
[39] D.K.Isenor, S.G.Zaky. Fingerprint identification using graph matching[J].Pattern Recognition,1986,19(2):113-122
[40] A.K.Hrechak, J.A.McHugh. Automated fingerprint recognition using structural matching. Pattern Recognition[J],1999,23(8):893-904
[41] Z.Chen, C.H.Kuo. A topology-based matching algorithm for fingerprint authentication[J]. Proceedings,25th Annual 1991 IEEE International Carnahan Conference on Security Technology, IEEE,1991:84-87
[42] B.G.Sherlock, D.M.Monro. A model for interpreting fingerprint topology[J]. Pattern Recognition[J],1993,26(7):1047-1055
[43]漆远,田捷,邓祥.基于遗传算法的指纹图匹配算法及应用[J].软件学报,2000,11(4):488-494
[44] Xiping Luo; Jie Tian; Yan Wu, A minutiae matching algorithm in fingerprint verification[J],IEEE,2000,4:.833-836
[45]詹小四,宁新宝,尹义龙等.一种改进的点模式指纹匹配方法[J].南京大学学报, 2003,39(4):491-498
[46] Tsai-Yang Jea, Venu Govindaraju: A minutia-based partial fingerprint recognition system[J]. Pattern Recognition 2005,38(10): 1672-1684
[47] A .Wahab, S.H. Chin, E.C. Tan,,Novel approach to automated fingerprint recognition[J],Image and Signal Processing, IEE Proceedings , 1998, 145(3):160-166,
[48]田捷,杨鑫.生物特征识别技术理论与应用[M],北京:电子工业出版社,2005:2-15
[49] Salil Prabhakar, Anil K. Jain, Sharath Pankanti. Learning fingerprint minutiae location and type[J]. Pattern Recognition,2002
[50]杨小冬等.自动指纹识别系统预处理技术及细节特征提取算法的研究[J],南京大学学报,2006, 42(4):352-360
[2]卢官明等.生物特征识别综述[J].南京邮电大学学报, 2007, 27(1):82~87
[3]彭章平.自动指纹识别系统研究[D].长沙:中南大学
[4]张新淼.基于改进的Gabor滤波器组指纹图像识别的研究和实现[D].天津:天津师范大学, 2009年3月
[5]付景广.指纹识别中若干关键算法的研究[D].北京:中国科学院软件研究所,2003
[6]汪龙峰.低质量指纹图像增强与形变指纹匹配的研究[D].长沙:中南大学, 2009
[7]张宇.指纹图像质量评价方法研究[D].济南:山东大学, 2009
[8]周晔华.指纹图像预处理算法研究[D].南京:南京理工大学,2008
[9]杨雪.指纹识别系统研究[D].哈尔滨:哈尔滨理工大学, 2009
[10]夏振华,石玉,于盛林.基于Gabor滤波器的指纹增强[J].工程图像学报, 2006(5):81~85
[11] B.G.Sherlock, D.M. Monro, K. Millard. Fingerprint enhancement by directional Fourier filtering[J]. IEE hoc.-Vis. Image Signal Process, 1994, 141(2):87~93
[12]傅景广,许刚,王裕国.基于二值图像的指纹细节点提取[J].计算机研究与发展,2004, 41(4):721-727
[13] Nalini IS. Ratha &Vinayaka D. Pandit. Robust Fingerprint Authentication Using Local Structural Similarity[J]. IEEE,2000:29-34
[14] M. M. Hadhoud, W. S. ElKilani, M. I. Samaan. An Adaptive Algorithm for Fingerprints Image Enhancement Using Gabor Filters[J]. IEEE , 2007:227-236
[15]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):855-862
[16] Lin Hong, Yifei Wan, and Anil Jain . Fingerprint Image Enhancement:Algorithm and Performance Evaluation. IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE, 1998, 20(8):777-789
[17]张建伟,陆亨立.指纹自动识别中图像分割方法的研究[J].微型电脑应用, 1999, 15(12) : 20-22
[18]黄贤武,苏鹏程,柏培权.基于方向滤波分割的指纹自动识别系统算法[J].中国图像图形报.2002 ,7 (8) :829.
[19]蒙丽彬,赵进创,傅文利.一种改进的基于G abor滤波的指纹增强算法[J].微计算机信息. 2006, 22(11-3):297-299
[20]叶四民,陈福祥.指纹图像预处理中的二值化技术[J].自动化与仪器仪表, 2001(2):30-39
[21]冯星奎,李林艳,颜祖泉.一种新的指纹图象细化算法[J].中国图像图形学报, 1999,4(10):835-837
[22]杨凡,赵顺东.一种有效的混合式指纹快速细化算法[J]. 2008, 25(10):3035-3041
[23]马宁.指纹图像的二值化与细化研究[D].南京:南京理工大学,2006
[24] Louisa Lam, Seong-Whan Lee. Thinning Methodologies-A Comprehensive Survey. IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE, 1992,14(9): 869-882
[25]王业琳,宁新宝,尹义龙.指纹图像细化算法的研究[J].南京大学学报. 2003, 39(4):468-475
[26] T.Y. Zhang & C.Y. Suen A fast Parallel thinning algorithm for thinning digital pattern. Communication of ACM.1984,27(3):236-239
[27]李小平,边肇祺,汪云九.二维Gabor滤波器的快速实现[J],自动化学报, 1989, 15(2):136-140
[28] Jan-Mark Geusebroek, Arnold W. M. Smeulders. Fast Anisotropic Gauss Filtering[J], IEEE TRANSACTIONS ON IMAGE PROCESSING,2003, 12(8):938-943
[29] Sanja Ranade, Azriel Rosenfeld. Point pattern matching by relaxation.Pattern Recognition[J],1980,12(4):269-275
[30] G.Stockman,S.Kopstein,S.Benett.Matching images to models for registration and object detection via clusterning[J] .IEEE Trans. On Pattern Analysis and Machine Intelligence,1982,4(3):229-241
[31] J.P.Pascual Starink, Eric Backer. Finding point correspondences using simulated annealing[J] .Pattern Recognition,1995,28(2):231-240
[32] A.K.Jain, Lin Hong, Sharath Pankanti, Ruud Bolle. An identity-anthentication system using fingerprints[J]. Proceedings of the IEEE,1997,85(9):1365-1388
[33]罗喜平,田捷.自动指纹识别总的图像增强和图像匹配[J].软件学报,2002,13(5):946-56
[34] Yuliang He, Jie Tian,Xiping Luo, Tanghui Zhang. Image enhancement and minutia matching in fingerprint verification[J].Pattern Recognition Letter,2003,24(9):1349-1360
[35] Xudong Jiang, Wei-Yun Yau. Fingerprint minutiae matching based on the local and global structures[J] .IEEE,2000:1038-1041
[36] C.I.Watson, P.J.Grother, D.P.Casasent. Distortion-tolerant fileter for elastic-distorted fingerprint matching[J].SPIE,2001
[37] Jain AK, Prabhakar S, Hong L, Pankanti S. Filterbank-based Fingerprint matching[J].IEEE Transaction On image Processing, 2000,9(5):846~859
[38] Anil Jain, Arun Ross, Salil Prabhakar. fingerprint matching using minutiae and texture features[J],in Proc. International Conference on Image Processing(ICIP), 2001:282-285
[39] D.K.Isenor, S.G.Zaky. Fingerprint identification using graph matching[J].Pattern Recognition,1986,19(2):113-122
[40] A.K.Hrechak, J.A.McHugh. Automated fingerprint recognition using structural matching. Pattern Recognition[J],1999,23(8):893-904
[41] Z.Chen, C.H.Kuo. A topology-based matching algorithm for fingerprint authentication[J]. Proceedings,25th Annual 1991 IEEE International Carnahan Conference on Security Technology, IEEE,1991:84-87
[42] B.G.Sherlock, D.M.Monro. A model for interpreting fingerprint topology[J]. Pattern Recognition[J],1993,26(7):1047-1055
[43]漆远,田捷,邓祥.基于遗传算法的指纹图匹配算法及应用[J].软件学报,2000,11(4):488-494
[44] Xiping Luo; Jie Tian; Yan Wu, A minutiae matching algorithm in fingerprint verification[J],IEEE,2000,4:.833-836
[45]詹小四,宁新宝,尹义龙等.一种改进的点模式指纹匹配方法[J].南京大学学报, 2003,39(4):491-498
[46] Tsai-Yang Jea, Venu Govindaraju: A minutia-based partial fingerprint recognition system[J]. Pattern Recognition 2005,38(10): 1672-1684
[47] A .Wahab, S.H. Chin, E.C. Tan,,Novel approach to automated fingerprint recognition[J],Image and Signal Processing, IEE Proceedings , 1998, 145(3):160-166,
[48]田捷,杨鑫.生物特征识别技术理论与应用[M],北京:电子工业出版社,2005:2-15
[49] Salil Prabhakar, Anil K. Jain, Sharath Pankanti. Learning fingerprint minutiae location and type[J]. Pattern Recognition,2002
[50]杨小冬等.自动指纹识别系统预处理技术及细节特征提取算法的研究[J],南京大学学报,2006, 42(4):352-360