指横纹识别中特征融合方法的研究
|
摘要
随着社会的进步和经济的发展，法制观念逐步加深，对个人身份识别的需求越来越广泛。鉴于传统身份识别方法中存在的一些弊端，生物特征识别技术开始受到人们的广泛关注。指横纹特征是一种新兴的生物特征，除了具有与其他手部特征一样的稳定性和可区分性特点外，还有其独有的特征。本文选择指横纹特征进行身份识别，并对识别过程中的特征提取算法展开研究。
在指横纹识别过程中，特征提取是一个关键的步骤，如何提取有效的特征就成为关注的重点问题之一。由于单一的生物特征进行身份识别有一定的局限性，信息融合被引入到指横纹识别中。本文对指横纹特征提取算法和特征级融合算法进行了深入的研究，主要工作如下：
(1)对指横纹的特征提取进行了研究。利用经典的特征提取方法，分别使用Gabor滤波提取指横纹特征、PCA提取指横纹特征，并提出了用局部二进制模式(LBP)的方法提取指横纹特征的方法。
(2)提出了基于Gabor与PCA相结合的指横纹特征提取的算法，首先对指横纹进行Gabor变换，然后使用PCA进行降维，最后根据欧氏距离对指横纹匹配识别。实验表明该方法得到的识别率要高于单独使用Gabor和PCA方法的结果。
(3)鉴于单一特征进行身份识别具有局限性，本文提出了特征融合的方法。应用Gabor滤波器对指横纹图像进行滤波得到其特征，然后融合四个手指的特征，实现了指横纹识别。
通过在北京交通大学指横纹数据库及实验室自行采集建立的图像库实验证明，本文算法可以有效地实现指横纹特征识别。
With the rapid development of modern science and technology, the concept of legalsystem gradually deepens, the demand for personal identification is more and more extensive,because of the drawbacks of the traditional identification methods, biometric technology wasborn, and is now widely used in various fields. Knuckle-print is an emerging biometric; it hasspecial features in addition to other characteristics such as the stability and distinguishcharacteristics. In this paper, the Knuckle-print was chosen for Biometric identification, andstudy the feature extraction algorithm in the identification process.
Feature extraction is the key to the whole process of the Knuckle-print recognition, howto extract features effectively is becoming one of the important attentions. Because of certainlimitations of single biological characteristics, information fusion is introduced to thebiological characteristics identification. This paper put forward feature extraction algorithmand character fusion algorithm, the main research work are as follows:
(1)Feature extraction of the Knuckle-print. We use the typical method of featureextraction which is the Gabor filter, PCA to extract features respectively, and propose amethod which is based on the local binary mode to extract the Knuckle-print feature.
(2) Put forward algorithm based on Gabor and PCA for the feature extraction of theKnuckle-print. The experimental results show that the recognition rate of the method is betterthan used Gabor and PCA method alone respectively.
(3) This paper puts forward method of the characteristics fusion for the Knuckle-printrecognition due to the limitations of the single feature identification.
The proposed algorithms can effectively implement knuckle-print identification. It isshown by experiments which proceed in our own building image data and the data providedby Information Institute of Beijing Jiaotong University.
在指横纹识别过程中，特征提取是一个关键的步骤，如何提取有效的特征就成为关注的重点问题之一。由于单一的生物特征进行身份识别有一定的局限性，信息融合被引入到指横纹识别中。本文对指横纹特征提取算法和特征级融合算法进行了深入的研究，主要工作如下：
(1)对指横纹的特征提取进行了研究。利用经典的特征提取方法，分别使用Gabor滤波提取指横纹特征、PCA提取指横纹特征，并提出了用局部二进制模式(LBP)的方法提取指横纹特征的方法。
(2)提出了基于Gabor与PCA相结合的指横纹特征提取的算法，首先对指横纹进行Gabor变换，然后使用PCA进行降维，最后根据欧氏距离对指横纹匹配识别。实验表明该方法得到的识别率要高于单独使用Gabor和PCA方法的结果。
(3)鉴于单一特征进行身份识别具有局限性，本文提出了特征融合的方法。应用Gabor滤波器对指横纹图像进行滤波得到其特征，然后融合四个手指的特征，实现了指横纹识别。
通过在北京交通大学指横纹数据库及实验室自行采集建立的图像库实验证明，本文算法可以有效地实现指横纹特征识别。
With the rapid development of modern science and technology, the concept of legalsystem gradually deepens, the demand for personal identification is more and more extensive,because of the drawbacks of the traditional identification methods, biometric technology wasborn, and is now widely used in various fields. Knuckle-print is an emerging biometric; it hasspecial features in addition to other characteristics such as the stability and distinguishcharacteristics. In this paper, the Knuckle-print was chosen for Biometric identification, andstudy the feature extraction algorithm in the identification process.
Feature extraction is the key to the whole process of the Knuckle-print recognition, howto extract features effectively is becoming one of the important attentions. Because of certainlimitations of single biological characteristics, information fusion is introduced to thebiological characteristics identification. This paper put forward feature extraction algorithmand character fusion algorithm, the main research work are as follows:
(1)Feature extraction of the Knuckle-print. We use the typical method of featureextraction which is the Gabor filter, PCA to extract features respectively, and propose amethod which is based on the local binary mode to extract the Knuckle-print feature.
(2) Put forward algorithm based on Gabor and PCA for the feature extraction of theKnuckle-print. The experimental results show that the recognition rate of the method is betterthan used Gabor and PCA method alone respectively.
(3) This paper puts forward method of the characteristics fusion for the Knuckle-printrecognition due to the limitations of the single feature identification.
The proposed algorithms can effectively implement knuckle-print identification. It isshown by experiments which proceed in our own building image data and the data providedby Information Institute of Beijing Jiaotong University.
引文
[1] Jain, R.Bole, S.Pankanti. Biometrics: Personal Identification in Networked Society [M]. KluwerAcademic Publishers.1999.1993，33（4）：62—67.
[2] The David Zhang. Automated Biometrics is one Technologies and Systems [M]. Kluwer AcademicPublishers.2000.
[3] Lawton G. Biometric Market and Industry Report2009-2014[DB/OL].http://www.biometricgroup.com/reports/public/market_report.php,2010.
[4] A.K.Jain, L.Hong, P.Harath, R.Bolle. An Identity-authentication System Using Fingerprints [J]. ProcIEEE.85（9）:1365-1388.
[5] K.Machida, S.Shigematsu, H.Morimura.Tanabe etc. A Novel Semiconductor Capacitive Sensor for aSingle-chip Fingerprint Sensor/Identifier LSI [J]. IEEE Transactions on Electron Devices.2001,48(10):2273-2278.
[6] Daugman J G. High confidence visual recognition of persons by a test of statistical independence [J].IEEE Trans. Pattern Anal and Maehine intell,1993,15(11):1148-1161.
[7] E.Newham. The Biometric Report.SJB Services [R].1995, New York.
[8] D.Maio, D.Mltoni, R. CppeLLi, J.L. Wayman, A.Jain, FVC2000: fingerprint Verification Competition[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence.2002,24(3):402-412.
[9] Li Q, Qiu Z, Sun. D, et al. Personal Identification Using Knuckle-print [J]. Advances in BiometricPerson Authentication,2004,3338:680-689.
[10]李强.手部特征识别及特征及融合算法研究[D].北京交通大学博士论文，2006.
[11]顾理，庄镇泉等.手部识别中的手形提取方法[J].计算机仿真，2005，22(7):128-132.
[12] Gu J, Zhou J, Yang C. Fingerprint Recognition by Combining Global Structure and Local Cues [J].IEEE Transactions on Image Processing,2006,15(7):1952-1964.
[13] Zhang L, Zhang L, Zhang D. Finger-knuckle-print Verification Based on Band-limited Phase-onlyCorrelation [J]. Computer Analysis of Image and Patterns,2009,5702:141-148.
[14]竺乐庆，张三元，幸锐.基于指节纹的个人身份自动识别[J].自动化学报，2009，35（7）：875-881.
[15] Rodriguez P, Silva J. Biometric identification by dermatoglyphics [J]. In: Proceedings of IEEEInternational Conference on Image Processing. IEEE,1996.312-322.
[16] S.Ribaric, I.Fratie. A Biometric Identification system Based on EigenPalm and Eigenfinger Features[J]. IEEE Trans on PAMI,2005，27(11):1698-170
[17]孙冬梅，掌纹与手形识别算法的研究[D].北京交通大学博士论文，2003.
[18]罗荣芳,林土胜,吴霆.基于人体手指指节折痕的身份识别方法[J].光学工程，2007,34(6):116-121.
[19]毛贤光,赖晓铮,赖声礼.一种新的指横纹与静脉加权融合算法[J].华南理工大学学报（自然科学版）,2009,37(1):74-78.
[20] Ribaric. S, Fratric. I, A Biometric Identification System Based on Eigenpalm and Eigenfinger Features[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence,2005,27(11):1698-1709.
[21] Nanni. L., Lumini. A. A multi-matcher system based on Knuckle-based features [J]. NeuralComputing&Applications,2009,18(1):87-91.
[22] Zhao R, Li. K.L, Liu. M, et al. A Novel Approach of Personal Identification Based on SingleKnuckle-print Image [J]. Asia-Pacific Conference on Information Processing,2009(2):218-221.
[23]李强,裘正定,孙冬梅,等.指横纹：一种新的生物身份特征[J].自动化学报.2007,33（6）：596-601.
[24] Shu W, Zhang D. Palmprint verification: an implementation of biometric technology [J]. In:Proceedings of the14th International Conference on Pattern Recognition. Brisbane, Australia: IEEE,1998.219-221.
[25] Zhang D, Shu W. Two novel characteristics in palmprint verification: datum point invariance and linefeature matching Pattern Recognition [J].1999,32(4):691-702.
[26] Li F, Leung M K H, Yu X Z. Palmprint matching using line features [J]. In: Proceedings of the8thInternational Conference on Advanced Communication Technology. Gangwon-Do, Korea: IEEE,2006.1577-1582.
[27] Li F, Leung M K H, Yu X Z. Palmprint identification using Hausdor distance [R]. In: Proceedings ofInternational Workshop on Biomedical Circuits and Systems. Singapore, Singapore: IEEE,2004:5-8.
[28] Lu G M, Zhang D, Wang K Q. Palmprint recognition using eigenpalms features. Pattern RecognitionLetters,2003,24(9-10):1463-1467.
[29] Wang M, Ruan Q Q. Palmprint recognition based on two-dimensional methods [C]. In: Proceedings ofthe8th International Conference on Signal Processing. Guilin, China: IEEE,2006.1-5.
[30] LiWX，ZhangD，Xu2Q. Palmprint Recognition Based on Fourier Transform.[J].2002Journal ofSoftware, Vol.13(5):236-238.
[31]苏晓生,林喜荣等.基于小波变幻的掌纹特征提取[J].清华大学学报(自然科学版)，2003，43(8):1049-1052.
[32] D.Gabor. Theory of communication. Journal of Institute for E1eetrieal Engineering [M].1946，93:429-457.
[33] Daugman J G. Complete Discrete2_D Gabor Transforms by Neutral of image Analysis andCompression [J]. IEEE, Trans.on Acoustics, Speech&Signal processing,1988,36(7):1169-1179.
[34] Daugman J G. Uncertainty Relation For Resolution in Space, Spatial Frequency and OrientationOptimized By Two-Dimensional Visual Cortical filters [J].1985,123(7):1160-1169.
[35] J.P.Jones, L.A Palmer. An evaluation of the two-dimensional Gabor filter model of simple receptivefields in cat striate cortex [J]. Journal of Neural physiology，1987,58(6):1233-1258.
[36]Turk M, Pentland A. Eigenfaces for Recognition [J]. Journal of Cognitive Neuroscience,1991,3(01):71-86.
[37] M. Topi, O. Timo, P. Matti, S. Maricor. Robust texture classification by subsets of local binarypatterns [C]. Proceedings of the15th International Conference on Pattern Recognition,2000,(3):935-938.
[38] G. Zhao, M. Pietikainen. Dynamic texture recognition using local binary patterns with an applicationto facial expressions [J]. IEEE Transaction on Pattern Analysis and Machine Intelligence,2007,29(6):915-928.
[39] T. Ahonen, A. Hadid, M. Pietikainen. Face description with local binary patterns: application to facerecognition [J]. IEEE Transaction on Pattern Analysis and Machine Intelligence,2006,28(12):2037-2041.
[40] Ojala T, Pletikainen M, Harwood D. A comparative study of texture measures with classificationbased on feature distributions [J]. Pattern Pecognition,1996.51-59.
[41] Ojala T, Pletikainen M, Multiresolution Gray-scale and Rotation Invariant Texture Classification withLocal Binary Patterns [J]. IEEE Transactions on Pattern Analysis and Machine Intelligence,2002.971-987.
[42] L. Hong, A. K. Jain, S.Pankanti. Can Multibiometric Improve Performance [NJ]. ProceedingsAutoID’99, Summit(NJ), USA,1999:59-64.
[43]杨万海.多传感器数据融合及其应用[M].西安电子科技大学出版社，2004.
[44] K.-A. TOh. Fingerprint and speaker verification decisions fusion [C].12th International Conferenceon Image Analysis and Processing，2003:626-631.
[45] C.Chen, C.T.Chu. Fusion of Face and Iris Features for Multimodal Biometrics [C]. InternationalConference on Biometric Autentication, HongKong:2006，LNCS3832:571-580
[46] S.Rabaric, D.Ribaric, N.Pavesic. A Biometric Identification System Based on the Fusion of Hand andPalm Features. Proceedings of The Advent of Biometrics on the Internet, A Cost275workshop, Rome,2002.
[47] A.Ross, A.K.Jain. Information Fusion in Biometrics [J].Pattern Recognition Letters,2003,24(13):2115-2125.
[48] J. Yang, J-y.Yang, D.Zhang, J-f.Lu. Feature Fusion: Parallel Strategy Vs. Serial Strategy [J]. PatternRecognition,2003,36(6):1369-1381.
[49] A.Kumar, D.Zhang. Biometric Recognition Using Feature Selection and Combination. AVBPA,2005:813-822.
[50] Y.Gao, M.Maggs. Feature-level Fusion in Personal Identification [C]. CVPR’05,2005:468-473.
[51] Wang Jiangang，YauWeiyun, Andy Suwandy and Eric Sung. Person recognition by fusing Palmprintand palm vein images based on "Laplacianpalm”, representation [J]. Patten Recognition.2008.41(5).1531-1544.
[52] FuYao，MaZhixing，QiMiao，LiJinsong，Li Xiaolu and LiuYinghua. A Novel User-Specific Face andPalmprint Feature level Fusion [J]. Proceedings of the2008Seeond International Symposium onIntelligent Information Technology Application.2008.3.296-300.
[2] The David Zhang. Automated Biometrics is one Technologies and Systems [M]. Kluwer AcademicPublishers.2000.
[3] Lawton G. Biometric Market and Industry Report2009-2014[DB/OL].http://www.biometricgroup.com/reports/public/market_report.php,2010.
[4] A.K.Jain, L.Hong, P.Harath, R.Bolle. An Identity-authentication System Using Fingerprints [J]. ProcIEEE.85（9）:1365-1388.
[5] K.Machida, S.Shigematsu, H.Morimura.Tanabe etc. A Novel Semiconductor Capacitive Sensor for aSingle-chip Fingerprint Sensor/Identifier LSI [J]. IEEE Transactions on Electron Devices.2001,48(10):2273-2278.
[6] Daugman J G. High confidence visual recognition of persons by a test of statistical independence [J].IEEE Trans. Pattern Anal and Maehine intell,1993,15(11):1148-1161.
[7] E.Newham. The Biometric Report.SJB Services [R].1995, New York.
[8] D.Maio, D.Mltoni, R. CppeLLi, J.L. Wayman, A.Jain, FVC2000: fingerprint Verification Competition[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence.2002,24(3):402-412.
[9] Li Q, Qiu Z, Sun. D, et al. Personal Identification Using Knuckle-print [J]. Advances in BiometricPerson Authentication,2004,3338:680-689.
[10]李强.手部特征识别及特征及融合算法研究[D].北京交通大学博士论文，2006.
[11]顾理，庄镇泉等.手部识别中的手形提取方法[J].计算机仿真，2005，22(7):128-132.
[12] Gu J, Zhou J, Yang C. Fingerprint Recognition by Combining Global Structure and Local Cues [J].IEEE Transactions on Image Processing,2006,15(7):1952-1964.
[13] Zhang L, Zhang L, Zhang D. Finger-knuckle-print Verification Based on Band-limited Phase-onlyCorrelation [J]. Computer Analysis of Image and Patterns,2009,5702:141-148.
[14]竺乐庆，张三元，幸锐.基于指节纹的个人身份自动识别[J].自动化学报，2009，35（7）：875-881.
[15] Rodriguez P, Silva J. Biometric identification by dermatoglyphics [J]. In: Proceedings of IEEEInternational Conference on Image Processing. IEEE,1996.312-322.
[16] S.Ribaric, I.Fratie. A Biometric Identification system Based on EigenPalm and Eigenfinger Features[J]. IEEE Trans on PAMI,2005，27(11):1698-170
[17]孙冬梅，掌纹与手形识别算法的研究[D].北京交通大学博士论文，2003.
[18]罗荣芳,林土胜,吴霆.基于人体手指指节折痕的身份识别方法[J].光学工程，2007,34(6):116-121.
[19]毛贤光,赖晓铮,赖声礼.一种新的指横纹与静脉加权融合算法[J].华南理工大学学报（自然科学版）,2009,37(1):74-78.
[20] Ribaric. S, Fratric. I, A Biometric Identification System Based on Eigenpalm and Eigenfinger Features[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence,2005,27(11):1698-1709.
[21] Nanni. L., Lumini. A. A multi-matcher system based on Knuckle-based features [J]. NeuralComputing&Applications,2009,18(1):87-91.
[22] Zhao R, Li. K.L, Liu. M, et al. A Novel Approach of Personal Identification Based on SingleKnuckle-print Image [J]. Asia-Pacific Conference on Information Processing,2009(2):218-221.
[23]李强,裘正定,孙冬梅,等.指横纹：一种新的生物身份特征[J].自动化学报.2007,33（6）：596-601.
[24] Shu W, Zhang D. Palmprint verification: an implementation of biometric technology [J]. In:Proceedings of the14th International Conference on Pattern Recognition. Brisbane, Australia: IEEE,1998.219-221.
[25] Zhang D, Shu W. Two novel characteristics in palmprint verification: datum point invariance and linefeature matching Pattern Recognition [J].1999,32(4):691-702.
[26] Li F, Leung M K H, Yu X Z. Palmprint matching using line features [J]. In: Proceedings of the8thInternational Conference on Advanced Communication Technology. Gangwon-Do, Korea: IEEE,2006.1577-1582.
[27] Li F, Leung M K H, Yu X Z. Palmprint identification using Hausdor distance [R]. In: Proceedings ofInternational Workshop on Biomedical Circuits and Systems. Singapore, Singapore: IEEE,2004:5-8.
[28] Lu G M, Zhang D, Wang K Q. Palmprint recognition using eigenpalms features. Pattern RecognitionLetters,2003,24(9-10):1463-1467.
[29] Wang M, Ruan Q Q. Palmprint recognition based on two-dimensional methods [C]. In: Proceedings ofthe8th International Conference on Signal Processing. Guilin, China: IEEE,2006.1-5.
[30] LiWX，ZhangD，Xu2Q. Palmprint Recognition Based on Fourier Transform.[J].2002Journal ofSoftware, Vol.13(5):236-238.
[31]苏晓生,林喜荣等.基于小波变幻的掌纹特征提取[J].清华大学学报(自然科学版)，2003，43(8):1049-1052.
[32] D.Gabor. Theory of communication. Journal of Institute for E1eetrieal Engineering [M].1946，93:429-457.
[33] Daugman J G. Complete Discrete2_D Gabor Transforms by Neutral of image Analysis andCompression [J]. IEEE, Trans.on Acoustics, Speech&Signal processing,1988,36(7):1169-1179.
[34] Daugman J G. Uncertainty Relation For Resolution in Space, Spatial Frequency and OrientationOptimized By Two-Dimensional Visual Cortical filters [J].1985,123(7):1160-1169.
[35] J.P.Jones, L.A Palmer. An evaluation of the two-dimensional Gabor filter model of simple receptivefields in cat striate cortex [J]. Journal of Neural physiology，1987,58(6):1233-1258.
[36]Turk M, Pentland A. Eigenfaces for Recognition [J]. Journal of Cognitive Neuroscience,1991,3(01):71-86.
[37] M. Topi, O. Timo, P. Matti, S. Maricor. Robust texture classification by subsets of local binarypatterns [C]. Proceedings of the15th International Conference on Pattern Recognition,2000,(3):935-938.
[38] G. Zhao, M. Pietikainen. Dynamic texture recognition using local binary patterns with an applicationto facial expressions [J]. IEEE Transaction on Pattern Analysis and Machine Intelligence,2007,29(6):915-928.
[39] T. Ahonen, A. Hadid, M. Pietikainen. Face description with local binary patterns: application to facerecognition [J]. IEEE Transaction on Pattern Analysis and Machine Intelligence,2006,28(12):2037-2041.
[40] Ojala T, Pletikainen M, Harwood D. A comparative study of texture measures with classificationbased on feature distributions [J]. Pattern Pecognition,1996.51-59.
[41] Ojala T, Pletikainen M, Multiresolution Gray-scale and Rotation Invariant Texture Classification withLocal Binary Patterns [J]. IEEE Transactions on Pattern Analysis and Machine Intelligence,2002.971-987.
[42] L. Hong, A. K. Jain, S.Pankanti. Can Multibiometric Improve Performance [NJ]. ProceedingsAutoID’99, Summit(NJ), USA,1999:59-64.
[43]杨万海.多传感器数据融合及其应用[M].西安电子科技大学出版社，2004.
[44] K.-A. TOh. Fingerprint and speaker verification decisions fusion [C].12th International Conferenceon Image Analysis and Processing，2003:626-631.
[45] C.Chen, C.T.Chu. Fusion of Face and Iris Features for Multimodal Biometrics [C]. InternationalConference on Biometric Autentication, HongKong:2006，LNCS3832:571-580
[46] S.Rabaric, D.Ribaric, N.Pavesic. A Biometric Identification System Based on the Fusion of Hand andPalm Features. Proceedings of The Advent of Biometrics on the Internet, A Cost275workshop, Rome,2002.
[47] A.Ross, A.K.Jain. Information Fusion in Biometrics [J].Pattern Recognition Letters,2003,24(13):2115-2125.
[48] J. Yang, J-y.Yang, D.Zhang, J-f.Lu. Feature Fusion: Parallel Strategy Vs. Serial Strategy [J]. PatternRecognition,2003,36(6):1369-1381.
[49] A.Kumar, D.Zhang. Biometric Recognition Using Feature Selection and Combination. AVBPA,2005:813-822.
[50] Y.Gao, M.Maggs. Feature-level Fusion in Personal Identification [C]. CVPR’05,2005:468-473.
[51] Wang Jiangang，YauWeiyun, Andy Suwandy and Eric Sung. Person recognition by fusing Palmprintand palm vein images based on "Laplacianpalm”, representation [J]. Patten Recognition.2008.41(5).1531-1544.
[52] FuYao，MaZhixing，QiMiao，LiJinsong，Li Xiaolu and LiuYinghua. A Novel User-Specific Face andPalmprint Feature level Fusion [J]. Proceedings of the2008Seeond International Symposium onIntelligent Information Technology Application.2008.3.296-300.