手背静脉识别算法研究
|
摘要
静脉识别是最近几年才开始发展的一种生物识别技术,它是利用每个人的手背血管分布不同这一特性进行身份认证的。与现在人们所熟知的指纹识别技术相比,具有更安全,非接触性等优点。
本文在收集和分析近年来国内外有关生物识别技术研究成果的基础上,对静脉识别系统的关键技术进行了研究。主要工作有以下几个方面:(1)静脉图像预处理。其目的是去除原始图像中的各种噪声,增加图像的清晰度,提取静脉的骨架结构,将图像变成只有单象素宽的点线图,以便于下一步的细节特征提取。具体包括静脉图像增强、二值化、滤波去噪、细化等几个部分。在静脉图像增强方面提出了频域增强后中值滤波、CLAHE(对比度受限的局部直方图均衡)和top-hat变换(顶帽变换)相结合的方法的手背静脉图像增强算法。实验结果表明,经过该算法处理后得到的图像得到显著增强,能够较好的反映静脉纹理。(2)静脉图像的特征提取。其目的是从预处理得到的静脉图像中,提取静脉的特征,包括图像的特征矩、所有端节点和分支点及其拓扑关系。文中用7个不变矩和小波矩来反映静脉图像的矩特征;用端节点和分支点的数目及线段特征来反映静脉几何结构;建立交叉点之间的拓扑关系,用最小二乘法对它们之间曲线形状进行拟合。最后运用信息融合技术将上述两组特征进行决策融合.提高了算法的整体识别性能,实验证明融合策略的效果最佳。
The hand vein recognition is a new biologiacl feature recognition wich is devoloepd only several year.We can carry on the boday identification through the band vein.Compared with fingerprint recognition,vein rccognition has many advantages, such as safely,non-contact.
Basde on the latest extensive discourse and tchnology journals in this field,this paper is trying to make studies on vein recognition system.Main works are as follows.
(1) Vein pre-plcoessin.Its purpose is to move noise of vein image and improving definition of vein image,get the structure of the vein hand,change vein image to a single pels width image,which will be beneficial to the following feature extraction.Pre-processing includes image enhancement,binarization,de-noising, thinning and so on.In the image enhancement,the improving median filter and the enhanced algorithm based on CLAHE and top-hat transform are statemented. Experimental results show that the image has been markedly enhanced after making use of these algorithms,and better reflected the vein texture.
(2) Vein feature extraction.Its purpose is to get all features from the pre-processing image,including image characteristic moment,ending and bifuecations and their topological relations.Seven invariant moments and wavelet moments were made to reflect the characteristics moment of hand vein image;the numbers of the ending and bifuecations and their segments demonsrated the geometry of hand vein;the topological relations of ending and bifuecations established,least square method was applied to fit the curve shape between them.Finally,the above-mentioned tow features decision-making integrated by employing information fusion technology.That improved the overall recognition performance of algorithm,and experimental results proved that the integration strategy was the best.
本文在收集和分析近年来国内外有关生物识别技术研究成果的基础上,对静脉识别系统的关键技术进行了研究。主要工作有以下几个方面:(1)静脉图像预处理。其目的是去除原始图像中的各种噪声,增加图像的清晰度,提取静脉的骨架结构,将图像变成只有单象素宽的点线图,以便于下一步的细节特征提取。具体包括静脉图像增强、二值化、滤波去噪、细化等几个部分。在静脉图像增强方面提出了频域增强后中值滤波、CLAHE(对比度受限的局部直方图均衡)和top-hat变换(顶帽变换)相结合的方法的手背静脉图像增强算法。实验结果表明,经过该算法处理后得到的图像得到显著增强,能够较好的反映静脉纹理。(2)静脉图像的特征提取。其目的是从预处理得到的静脉图像中,提取静脉的特征,包括图像的特征矩、所有端节点和分支点及其拓扑关系。文中用7个不变矩和小波矩来反映静脉图像的矩特征;用端节点和分支点的数目及线段特征来反映静脉几何结构;建立交叉点之间的拓扑关系,用最小二乘法对它们之间曲线形状进行拟合。最后运用信息融合技术将上述两组特征进行决策融合.提高了算法的整体识别性能,实验证明融合策略的效果最佳。
The hand vein recognition is a new biologiacl feature recognition wich is devoloepd only several year.We can carry on the boday identification through the band vein.Compared with fingerprint recognition,vein rccognition has many advantages, such as safely,non-contact.
Basde on the latest extensive discourse and tchnology journals in this field,this paper is trying to make studies on vein recognition system.Main works are as follows.
(1) Vein pre-plcoessin.Its purpose is to move noise of vein image and improving definition of vein image,get the structure of the vein hand,change vein image to a single pels width image,which will be beneficial to the following feature extraction.Pre-processing includes image enhancement,binarization,de-noising, thinning and so on.In the image enhancement,the improving median filter and the enhanced algorithm based on CLAHE and top-hat transform are statemented. Experimental results show that the image has been markedly enhanced after making use of these algorithms,and better reflected the vein texture.
(2) Vein feature extraction.Its purpose is to get all features from the pre-processing image,including image characteristic moment,ending and bifuecations and their topological relations.Seven invariant moments and wavelet moments were made to reflect the characteristics moment of hand vein image;the numbers of the ending and bifuecations and their segments demonsrated the geometry of hand vein;the topological relations of ending and bifuecations established,least square method was applied to fit the curve shape between them.Finally,the above-mentioned tow features decision-making integrated by employing information fusion technology.That improved the overall recognition performance of algorithm,and experimental results proved that the integration strategy was the best.
引文
[1]林喜荣.人体手背血管图像的特征提取及匹配.清华大学学报(自然科学版),2003,43(2):164-167.
[2]王科俊丁宇航,王大振.基于静脉识别的身份认证方法研究.科技导报,2005,23(1):35-37.
[3]崔国范,温学兵.人体静脉识别技术概述[J].中国科技信息.2007,16.
[4]Paranjape RB.Adaptive neighborhood histogram equalization for image enhancement[J].CVGIP,1992,54(3):259-267.
[5]Pizer SM,Amburn EP.Adap tive histogram eqalization and its variations[J].GVGIP,1987,39:355-368.
[6]Kim TK,Kim LS,Hwang SH.An advanced contrast enhancement using partially overlapped sub - black histogram equalization[J].IEEE ISCAS,2000,11(6):236 - 250,Geneva,Switzerland,2000 -2005.
[7]Kim TK,Kim LS,Hwang SH.An advanced contrast enhancement using partially overlapped sub - black histogram equalization[J].IEEE ISCAS,2000,11(6):236 - 250,Geneva,Switzerland,2000 -2005.
[8]Barnes JE.Characteristics and control of contrast in CT[J].Radi -ographics,1992,12(3):825 - 837.
[9]Barnes JE.Characteristics and control of contrast in CT[J].Radi -ographics,1992,12(3):825 - 837.
[10]Barnes JE.Characteristics and control of contrast in CT[J].Radi-ographics,1992,12(3):825 - 837.
[11]冈萨雷斯.数字图像处理.[M].北京:电子工业出版社,2005,9.
[12]金晨望,郭佑民,强永乾.基于直方图均衡的DICOM图像增强显示技术初步研究[J].影像技术学2006,22(4):466-469
[13]徐力平,蔡艳艳.基于CLAHE的尘肺X线胸片增强技术[J].计算机应用.2007.27:388-389
[14]戴青云,余英林.数学形态学在图像处理中的应用进展.[J].控制理论与应用.2001,18(4):478-482
[15]江萍,徐晓冰,方敏.基于top-hat变换的车牌快速搜索与定位[J].合肥工业大学学报.2003,26(3):336-339
[16]宣航,齐敏,M.Lehmpfuhl.一种基于白top-hat变换的医学图像的局部对比度增强算法[EB/OL].科技论文在线
[17]沈庭芝,方子文.数字图像处理及模式识别.1998,北京:北京理工大学出版社.
[18]章毓晋,图像工程(上册):图象处理和分析.1999,北京:清华大学出版社.
[19]阮秋奇,数字图像处理学.2001,北京:电子工业出版社.
[20]周密,李尊尊,耿国华.基于小波阈值的图像去噪方法研究[J].计算机技术与发展.2008.18(5):22-24
[21]刘文萍,吴立德.一种对空中目标自适应分割方法[J].红外与毫米波学报,1996,15(4):257-261
[22]涂其远,吴建华,万国金.动态阈值结合全局阈值对图像进行分割[J].南昌大学学报(工科版),2002,24(1):37-40.
[23]陈果,左洪福.图像阈值分割两种新技术[J].模式识别与人工智能,2002,15(4):468-472
[24]王科俊,丁宇航,庄大燕.手背静脉阈值图像分割[J].控制理论与应用,2005,24(8):19-22
[25]Hongwei Zhu,Otman Basir.Proximity measure image based region merging for texture segmentation through gabor filtering and watershed transform[C].Proceedings IEEE International Conference on Intelligent Systems and Signal Processing,2003(2):742-747.
[26]Paul R Hill,Nishan Canagarajah C,David R Bull.Image segment action using a texture gradient based watershed transform[J].IEEE Transactionson Image Processing,2003,12(12):1618 - 1633.
[27]裘旭光,张大鹏,刘晶.基于自适应区域增长算法获取手部静脉模式[J].计算机工程与应用.2007,43(33):230-233
[28]OtsuN.IEEE Trans SystcmMan Cybernct,1979,SMC.9:62-66P
[29]刘文萍,吴立德.一种对空中目标自适应分割方法[J].红外与毫米波学报.1996.15(4):257-261
[30]付小宁,殷世民,吴志鹏.红外图像的动态阈值分割[J].光电工程202,29(6):57-6
[31]刘秀兰,马丹,刘兵一种新的快速动态阈值图像分割算法[J]京工业大学学报.1999,25(2):92-95
[32]陈果.图像阈值分割的Fisher准则函数法[J].仪器仪表学报.203,24(6):564-567,57
[33]OuoGan;Moschytz,GS.Finger Print feature matching using CNNs.Circuit sand systems,
[34]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(自然科学), 2003,39(4):468-475.
[35]王大振,王科俊.手背静脉识别的关键技术研究[D].哈尔滨:哈尔滨工程大学,2005.
[36]秦筱,蔡超,周成千.一种有效的骨架毛刺去除算法[J].华中科技大学学报(自然科学版),2004,32(12):28-31.
[37]张晋阳,孙懋珩.手背静脉图像骨架特征提取的算法[J].计算机应用.2007,27(1):152-154
[38]E.Zhu,J.P.Yin,G.M,Zhang,Fingerprint matching based on global alignment of multiple reference minutiae,Pattern recognition 38(10)(2005) 1685-1694P
[39]M.Kawagoe,A.Tojo,Fingerprint Pattern classification,Pattern Recognition 17(3)(1984)295-303P
[40]刘文萍,吴立德一种对空中目标自适应分割方法[J].红外与毫米波学报.1996.15(4):257-261页
[41]韩学东.正面人脸图像识别系统研究与实现.哈尔滨工程大学硕士学位论文.2000:21-25
[42]Y.L.Sheng,H.H.Arsenault.Noisy-image normalization using low-order radial moments of circular-harmonic function[J].J.Opt.Soc.Am.A,1987,4(7):1176-1184P
[43]B.Wang,C.C.Sun.Enhancement of signal-to-noise ratio of a double random phase encoding encryption system[J].OPt.Eng.2001,40(8):1502-1506P
[44]袁华,董守斌,张凌等.GFO和不变矩实现基于形状的图像检索新方法[J].华南理工大学学报.2002,30(4):60-64页
[45]金丰华,秦磊,汪蕙等.几何矩不变量在基于内容医学图像检索中的应用[J].山东生物医学工程.2002,21(2):8-20页
[46]H.P.Ren,Z.L.Ping,W.R.GBO.Cell Image Recognition With Radial Harmonic Fourier Moment[J].Chinese physics,2003,12(6):41-62p
[47]Hu M K.Visual.Pattern Recognition by Moment Invariants[J].IEE Trans.Information Theory,1962,8:179-187.
[48]Wong Y.R.Scene Matching with Invariant Moments[J].Computer Graphics and lmageProcessing,1978,8:16-24.
[49]丁明跃.不变矩算法研究[J].数据采集与处理,1992,7(1):1-9.
[50]金丰华,秦磊,汪蕙,罗立民.几何矩不变量在基于内容医学图像检索中的应用[z].山东生物医学工程.2(X)2,21(2):7-10
[51]金琪,戴汝为.基于矩表示的小波不变量.模式识别与人工智能.1996,8(3):20-50
[52]杨万海.多传感器数据融合及其应用[M].西安:西安电子科技大学出版社,2004
[53]彭真明,张启衡,魏宇星等.基于多特征融合的图像匹配模式[J].强激光与粒子束.2004,16(3):281-285
[54]刘宏,刘宏,李锦涛等.多方法融合来解决人脸检测中的光照补偿[J].系统仿真学报.2001,13(增):486-489
[55]廖闻剑,成瑜.基于多分类器决策融合的印鉴真伪鉴别方法[J].南京航空航天大学学报.2002:34(4):368-371
[56]Teti J G.Gorman R P,Berger W A.A multifeature decision space approach to radar target identification[J].IEEE Trans.On AES.,1996,32(1):480-486
[57]顾理,庄镇泉,郑光勇,王再见.基于特征融合的手形匹配算法.计算机应用.2005年25卷10期:2286-2288
[2]王科俊丁宇航,王大振.基于静脉识别的身份认证方法研究.科技导报,2005,23(1):35-37.
[3]崔国范,温学兵.人体静脉识别技术概述[J].中国科技信息.2007,16.
[4]Paranjape RB.Adaptive neighborhood histogram equalization for image enhancement[J].CVGIP,1992,54(3):259-267.
[5]Pizer SM,Amburn EP.Adap tive histogram eqalization and its variations[J].GVGIP,1987,39:355-368.
[6]Kim TK,Kim LS,Hwang SH.An advanced contrast enhancement using partially overlapped sub - black histogram equalization[J].IEEE ISCAS,2000,11(6):236 - 250,Geneva,Switzerland,2000 -2005.
[7]Kim TK,Kim LS,Hwang SH.An advanced contrast enhancement using partially overlapped sub - black histogram equalization[J].IEEE ISCAS,2000,11(6):236 - 250,Geneva,Switzerland,2000 -2005.
[8]Barnes JE.Characteristics and control of contrast in CT[J].Radi -ographics,1992,12(3):825 - 837.
[9]Barnes JE.Characteristics and control of contrast in CT[J].Radi -ographics,1992,12(3):825 - 837.
[10]Barnes JE.Characteristics and control of contrast in CT[J].Radi-ographics,1992,12(3):825 - 837.
[11]冈萨雷斯.数字图像处理.[M].北京:电子工业出版社,2005,9.
[12]金晨望,郭佑民,强永乾.基于直方图均衡的DICOM图像增强显示技术初步研究[J].影像技术学2006,22(4):466-469
[13]徐力平,蔡艳艳.基于CLAHE的尘肺X线胸片增强技术[J].计算机应用.2007.27:388-389
[14]戴青云,余英林.数学形态学在图像处理中的应用进展.[J].控制理论与应用.2001,18(4):478-482
[15]江萍,徐晓冰,方敏.基于top-hat变换的车牌快速搜索与定位[J].合肥工业大学学报.2003,26(3):336-339
[16]宣航,齐敏,M.Lehmpfuhl.一种基于白top-hat变换的医学图像的局部对比度增强算法[EB/OL].科技论文在线
[17]沈庭芝,方子文.数字图像处理及模式识别.1998,北京:北京理工大学出版社.
[18]章毓晋,图像工程(上册):图象处理和分析.1999,北京:清华大学出版社.
[19]阮秋奇,数字图像处理学.2001,北京:电子工业出版社.
[20]周密,李尊尊,耿国华.基于小波阈值的图像去噪方法研究[J].计算机技术与发展.2008.18(5):22-24
[21]刘文萍,吴立德.一种对空中目标自适应分割方法[J].红外与毫米波学报,1996,15(4):257-261
[22]涂其远,吴建华,万国金.动态阈值结合全局阈值对图像进行分割[J].南昌大学学报(工科版),2002,24(1):37-40.
[23]陈果,左洪福.图像阈值分割两种新技术[J].模式识别与人工智能,2002,15(4):468-472
[24]王科俊,丁宇航,庄大燕.手背静脉阈值图像分割[J].控制理论与应用,2005,24(8):19-22
[25]Hongwei Zhu,Otman Basir.Proximity measure image based region merging for texture segmentation through gabor filtering and watershed transform[C].Proceedings IEEE International Conference on Intelligent Systems and Signal Processing,2003(2):742-747.
[26]Paul R Hill,Nishan Canagarajah C,David R Bull.Image segment action using a texture gradient based watershed transform[J].IEEE Transactionson Image Processing,2003,12(12):1618 - 1633.
[27]裘旭光,张大鹏,刘晶.基于自适应区域增长算法获取手部静脉模式[J].计算机工程与应用.2007,43(33):230-233
[28]OtsuN.IEEE Trans SystcmMan Cybernct,1979,SMC.9:62-66P
[29]刘文萍,吴立德.一种对空中目标自适应分割方法[J].红外与毫米波学报.1996.15(4):257-261
[30]付小宁,殷世民,吴志鹏.红外图像的动态阈值分割[J].光电工程202,29(6):57-6
[31]刘秀兰,马丹,刘兵一种新的快速动态阈值图像分割算法[J]京工业大学学报.1999,25(2):92-95
[32]陈果.图像阈值分割的Fisher准则函数法[J].仪器仪表学报.203,24(6):564-567,57
[33]OuoGan;Moschytz,GS.Finger Print feature matching using CNNs.Circuit sand systems,
[34]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(自然科学), 2003,39(4):468-475.
[35]王大振,王科俊.手背静脉识别的关键技术研究[D].哈尔滨:哈尔滨工程大学,2005.
[36]秦筱,蔡超,周成千.一种有效的骨架毛刺去除算法[J].华中科技大学学报(自然科学版),2004,32(12):28-31.
[37]张晋阳,孙懋珩.手背静脉图像骨架特征提取的算法[J].计算机应用.2007,27(1):152-154
[38]E.Zhu,J.P.Yin,G.M,Zhang,Fingerprint matching based on global alignment of multiple reference minutiae,Pattern recognition 38(10)(2005) 1685-1694P
[39]M.Kawagoe,A.Tojo,Fingerprint Pattern classification,Pattern Recognition 17(3)(1984)295-303P
[40]刘文萍,吴立德一种对空中目标自适应分割方法[J].红外与毫米波学报.1996.15(4):257-261页
[41]韩学东.正面人脸图像识别系统研究与实现.哈尔滨工程大学硕士学位论文.2000:21-25
[42]Y.L.Sheng,H.H.Arsenault.Noisy-image normalization using low-order radial moments of circular-harmonic function[J].J.Opt.Soc.Am.A,1987,4(7):1176-1184P
[43]B.Wang,C.C.Sun.Enhancement of signal-to-noise ratio of a double random phase encoding encryption system[J].OPt.Eng.2001,40(8):1502-1506P
[44]袁华,董守斌,张凌等.GFO和不变矩实现基于形状的图像检索新方法[J].华南理工大学学报.2002,30(4):60-64页
[45]金丰华,秦磊,汪蕙等.几何矩不变量在基于内容医学图像检索中的应用[J].山东生物医学工程.2002,21(2):8-20页
[46]H.P.Ren,Z.L.Ping,W.R.GBO.Cell Image Recognition With Radial Harmonic Fourier Moment[J].Chinese physics,2003,12(6):41-62p
[47]Hu M K.Visual.Pattern Recognition by Moment Invariants[J].IEE Trans.Information Theory,1962,8:179-187.
[48]Wong Y.R.Scene Matching with Invariant Moments[J].Computer Graphics and lmageProcessing,1978,8:16-24.
[49]丁明跃.不变矩算法研究[J].数据采集与处理,1992,7(1):1-9.
[50]金丰华,秦磊,汪蕙,罗立民.几何矩不变量在基于内容医学图像检索中的应用[z].山东生物医学工程.2(X)2,21(2):7-10
[51]金琪,戴汝为.基于矩表示的小波不变量.模式识别与人工智能.1996,8(3):20-50
[52]杨万海.多传感器数据融合及其应用[M].西安:西安电子科技大学出版社,2004
[53]彭真明,张启衡,魏宇星等.基于多特征融合的图像匹配模式[J].强激光与粒子束.2004,16(3):281-285
[54]刘宏,刘宏,李锦涛等.多方法融合来解决人脸检测中的光照补偿[J].系统仿真学报.2001,13(增):486-489
[55]廖闻剑,成瑜.基于多分类器决策融合的印鉴真伪鉴别方法[J].南京航空航天大学学报.2002:34(4):368-371
[56]Teti J G.Gorman R P,Berger W A.A multifeature decision space approach to radar target identification[J].IEEE Trans.On AES.,1996,32(1):480-486
[57]顾理,庄镇泉,郑光勇,王再见.基于特征融合的手形匹配算法.计算机应用.2005年25卷10期:2286-2288