基于数学形态学的人脸检测研究
|
摘要
随着智能化信息技术的发展,远程教育、视频监控、人机交互技术以及安全等各领域都迫切希望能够进行高效、准确的身份验证。人脸识别是一种根据人的眼睛、嘴巴等面部特征来自动进行身份识别的生物特征鉴别技术。人脸识别与其它生物特征识别技术相比较有其独特的优点,尤其是直观、非侵扰、防欺诈、良好的可扩展性等特点,决定了其具有广泛的应用领域,在国际学术界和工业界受到高度的关注,吸引了越来越多的研究者的兴趣,并在一些领域得到成功的应用,取得了一批有意义的研究成果。
由于人脸区域颜色信息丰富,而且易受光线等外界的干扰,所以人脸特征在不同姿态、不同光照或其他成像条件下,会出现多种不同的情况,增加了人脸检测难度。所以人脸识别一般存在计算复杂、速度慢、鲁棒性不佳等特点。本文综述了近年来国内外大量的关于人脸识别的文献和研究报告,在此基础上提出了基于数学形态学的人脸识别与特征检测方法。
首先,对人脸检测问题分类以及发展现状做了介绍,并通过对各种常见颜色空间进行一一的分析和比较,论述了YCbCr空间作为肤色建模的优越性,传统方式中忽略了肤色在YCbCr空间中Y分量的特性,仅考虑色度分量Cb和Cr。这种肤色模型存在的缺陷是:在图像的高亮度区和低亮度区之间,肤色色度才是基本不受亮度影响,所以要得到更准确的检测效果,必须考虑Y值不同而造成的影响。本文在统计了大量亮度不同人脸图像在YCbCr空间中三个分量的值,对肤色进行分段阈值判断,细化了肤色在不同亮度条件下的Cb,Cr所取的阈值范围,优化并改进了肤色模型。其次,人脸区域确定后,为了确定人眼和嘴巴特征位置,我们从数学形态学方法入手,先是利用数学形态学方法具有滤波的功能,设计了数学形态学滤波器,得到完整的人脸区域,再利用人眼和嘴巴在二值化图像中的形状,根据人眼和嘴巴特征结构在二值图像中的面积大小,设计了自适应可改变大小分区域的数学形态学模板检测算法,精确定位了人眼和嘴巴的位置。最后据此设计并实现了一个彩色图像人脸识别系统。实验结果表明,本文的方法,能够对人脸上的眼睛和嘴巴特征点进行快速和准确的定位,并且程序运行时间较短,具有一定的实时性,具有一定的理论意义和应用价值。
With the development of intellectualized information technology,high effective and accuracy Authentication is need in remote education,video monitoring,interactive technology, and safety areas.Face recognition technology is a biometric identification technology which identifies the status automatically on the basis of the features of human faces such as eyes and mouth.Face recognition technology has a lot of unique advantages among various biometric identification technologies,especially the advantages of visual、non-infestation、cheat-proof and good expansibility,which make it has a broad application domain.It also draws extensive attention in academic and the industrial circles and has been successfully applied in some fields. It also attracts more and more researchers' interesting and they have achieved some meaningful research achievements.
Because the color information in the person face region is rich,and moreover easily disturbed by outside light and other factors,then the person face characteristics will have many kinds of different situations under the different posture,the different illumination or other image formation condition,which increased the person face examination difficulty.The person face recognition and the feature detection general exists many characteristics such as complex computation,slow speed,bad robustness and so on.This paper reviews the recent years' massive domestic and foreign literature and articles about face recognition.Based on the massive study, this paper proposed the face recognition method based on mathematics morphology.
Firstly,it introduced the person face detection question's classification as well as the present development situation,and then elaborated the YCbCr space's superiority to take the skin color model by analyzing and comparing each kind of common color space.It has neglected the skin color in the YCbCr space the Y component characteristic,only considers chromaticity component Cb and Cr in the traditional way,the skin color model has such flows:between the high luminance area and the low brightness area,the skin color chromaticity is not basic brightness influence,that is to say,to obtain the more accurate examination effect,the Y value different creates influence should be considered.This paper has counted the massive brightness different person face image's three component values in the YCbCr space,carried on the partition threshold value judgment to the skin color,refined the threshold value of scope skin color which took under different brightness condition's Cb and Cr,optimized and improved the skin color model.Secondly,to determine human eye and the mouth characteristic position after person face region's determination,we obtained from mathematics morphology method,first designed mathematics morphology filter to obtain the complete person face region by using mathematics morphology's filter function,then designed the region mathematics morphology template examination algorithm which could change the divided size automatically,to pinpoint human eye and mouth's position through the human eye and mouth's shape in binary image, mathematics morphology template principle and according to the human eye and the mouth characteristic structure in binary image's area size.Finally it designed and realized a face detection and features location system in color image.The experimental results show that the method of the paper can location the human's eyes and mouth fast and accurate,the program running time is short,it has a certain real-time character,and it has a certain theoretical meaning and practical value.
由于人脸区域颜色信息丰富,而且易受光线等外界的干扰,所以人脸特征在不同姿态、不同光照或其他成像条件下,会出现多种不同的情况,增加了人脸检测难度。所以人脸识别一般存在计算复杂、速度慢、鲁棒性不佳等特点。本文综述了近年来国内外大量的关于人脸识别的文献和研究报告,在此基础上提出了基于数学形态学的人脸识别与特征检测方法。
首先,对人脸检测问题分类以及发展现状做了介绍,并通过对各种常见颜色空间进行一一的分析和比较,论述了YCbCr空间作为肤色建模的优越性,传统方式中忽略了肤色在YCbCr空间中Y分量的特性,仅考虑色度分量Cb和Cr。这种肤色模型存在的缺陷是:在图像的高亮度区和低亮度区之间,肤色色度才是基本不受亮度影响,所以要得到更准确的检测效果,必须考虑Y值不同而造成的影响。本文在统计了大量亮度不同人脸图像在YCbCr空间中三个分量的值,对肤色进行分段阈值判断,细化了肤色在不同亮度条件下的Cb,Cr所取的阈值范围,优化并改进了肤色模型。其次,人脸区域确定后,为了确定人眼和嘴巴特征位置,我们从数学形态学方法入手,先是利用数学形态学方法具有滤波的功能,设计了数学形态学滤波器,得到完整的人脸区域,再利用人眼和嘴巴在二值化图像中的形状,根据人眼和嘴巴特征结构在二值图像中的面积大小,设计了自适应可改变大小分区域的数学形态学模板检测算法,精确定位了人眼和嘴巴的位置。最后据此设计并实现了一个彩色图像人脸识别系统。实验结果表明,本文的方法,能够对人脸上的眼睛和嘴巴特征点进行快速和准确的定位,并且程序运行时间较短,具有一定的实时性,具有一定的理论意义和应用价值。
With the development of intellectualized information technology,high effective and accuracy Authentication is need in remote education,video monitoring,interactive technology, and safety areas.Face recognition technology is a biometric identification technology which identifies the status automatically on the basis of the features of human faces such as eyes and mouth.Face recognition technology has a lot of unique advantages among various biometric identification technologies,especially the advantages of visual、non-infestation、cheat-proof and good expansibility,which make it has a broad application domain.It also draws extensive attention in academic and the industrial circles and has been successfully applied in some fields. It also attracts more and more researchers' interesting and they have achieved some meaningful research achievements.
Because the color information in the person face region is rich,and moreover easily disturbed by outside light and other factors,then the person face characteristics will have many kinds of different situations under the different posture,the different illumination or other image formation condition,which increased the person face examination difficulty.The person face recognition and the feature detection general exists many characteristics such as complex computation,slow speed,bad robustness and so on.This paper reviews the recent years' massive domestic and foreign literature and articles about face recognition.Based on the massive study, this paper proposed the face recognition method based on mathematics morphology.
Firstly,it introduced the person face detection question's classification as well as the present development situation,and then elaborated the YCbCr space's superiority to take the skin color model by analyzing and comparing each kind of common color space.It has neglected the skin color in the YCbCr space the Y component characteristic,only considers chromaticity component Cb and Cr in the traditional way,the skin color model has such flows:between the high luminance area and the low brightness area,the skin color chromaticity is not basic brightness influence,that is to say,to obtain the more accurate examination effect,the Y value different creates influence should be considered.This paper has counted the massive brightness different person face image's three component values in the YCbCr space,carried on the partition threshold value judgment to the skin color,refined the threshold value of scope skin color which took under different brightness condition's Cb and Cr,optimized and improved the skin color model.Secondly,to determine human eye and the mouth characteristic position after person face region's determination,we obtained from mathematics morphology method,first designed mathematics morphology filter to obtain the complete person face region by using mathematics morphology's filter function,then designed the region mathematics morphology template examination algorithm which could change the divided size automatically,to pinpoint human eye and mouth's position through the human eye and mouth's shape in binary image, mathematics morphology template principle and according to the human eye and the mouth characteristic structure in binary image's area size.Finally it designed and realized a face detection and features location system in color image.The experimental results show that the method of the paper can location the human's eyes and mouth fast and accurate,the program running time is short,it has a certain real-time character,and it has a certain theoretical meaning and practical value.
引文
[1]Mark Pickering,Adrian Whichello,Michael Frater,John Arnold.A proposal for an automatic face extraction algorithm,ISO/IEC JTC1/SC29/WG11,MPEG99/5399,December 1999.
[2]郑庆．基于肤色的人脸检查与人眼定位[D]．电子科技大学,2007．5-6,25-26．
[3]Erik Hjelmas,Boon Kee Low.Face detection:a survey computer vision and image understanding,2001,83:236-274.
[4]M.Turk,Pentland.Eigenfaces for recognition.J.of cognitive Neuroscience,1991,3(1):71-76.
[5]N.Intrator,D.Reisfeld,Y.Yeshurun.Face recognition using a hybrid supervised unsupervised neural-network,PRL (17),January 1996,No.Ⅰ.:67-76.
[6]H.A.Rowley,S.Baluja,T.kanade.Rotation invariant neural network-based face detection.IEEE Transaction on Pattern Analysis and Machine Intelligence.1998.
[7]E.Osuna,R.Freund,F.Girosi.Training support vector machines:an application to face detection.Proc.IEEE Conf.on Computer Vision and Pattern Recognition.June 1998.
[8]N.Littleston.Learning quickly when irrelevant attributes bound:a new linear-threshold algorithm.Machine learning,1998.Vol.2:285-278.
[9]Rainer Lienhart and Jochen Maydt.An extended set of haar-like features for rapid object detection.IEEE ICIP 2002,Sep.2002.Vol.1:900-903.
[10]Paul Viola,Michael Jones.Rapid object detection using a boosted cascade of simple features.IEEE Computer Vision and Pattern Recognition (CVPRV'01).2001.Vol.l:511-518.
[11]P.Belhummer,J.Hespanha,D.Kriegman.Eigenfaces versus fisher-faces:recognition using class linear projection.IEEE Trans,on Pattern Analysis and machine Intelligence.1997.19(7):711-720.
[12]F.Samaria,S.Young.HMM based architecture for face identification.Image and Vision Computing.1994.Vol.(12):537-583.
[13]K-K Sung.Learning and example selection for object and pattern detection,PhD Thesis,Massachusetts Institute of Technology.1996.
[14]AN Rajagopalan,KS Kumar,J.Karlekar,R.Manivasakan,and MM Patil.Finding faces in photographs.In Proceedings of the Sixth International Conference on Computer Vision.1998.
[15]M.H.Yang,N.Ahuja.Detection human Faces in Color Images.In IEEE International Conference on Image Processing,Chicago,IL,October 4-7 1998:127-130.
[16]E.Saber,A.M.Tekalp.Frontal-view face detection and facial feature extraction using color,shape and symmetry based cost functions.Pattern Recognition Letters.1998.19(8): 669-680.
[17]J.Cai,A.Goshasby.Detecting human faces in color images.Image and Vision Computing.1999.18:63-74.
[18]王文宁,王汇源。彩色图像中多姿态的人脸快速检测与定位[J],计算机工程,2005,31(7):149-151.
[19]徐战武,朱森良.基于颜色的皮肤检测综述[J],中国图形图像学报,2007,3(12):203-209。
[20]彭波,崔永普,黄丹霞等.基于肤色和唇色信息的人脸检测方法的研究[J],计算机工程与设计,2006,26(6):1500-1502.
[21]陈泽宇,戚飞虎,陈刚.利用颜色信息的人脸检测方法[[J],上海交通大学学报,2000,34(6):793-795.
[22]党治,冯晓毅.基于新的肤色模型的人脸检测方法[J],计算机应用,2006,26(3):615-617
[23]Haralick R,Zhuang X.Image analysis using mathematical morphology.IEEE Trans.On Pattern Analysis and Machine Intelligence.1997,9(4):532-550.
[24]崔屹,图像处理与分析-数理形态学方法及应用[M],北京科学出版社,2000。
[25]Matheron G..Random Sets and Integral Geometry[M].New York,1995.
[26]T Martin,Hagan Howard,Mark Demuth,H.Beale,神经网络设计[M].机械工业出版社,2002,9。
[27]Tom M,Mitchell。机器学习[M],北京,机械工业出版社,2003.
[28]阮秋琦,数字图像处理学[M],电子工业出版社,2001。
[29]董长虹,赖志国,余啸海,MATLAB图像处理与应用[M],国防工业出版社,2004:182-195
[30]Feng G.C,Yuen P.C.,Variance projection function and its application to eye detection for human face recognition[J],Pattern Recognition Letters,1998,19:899-906.
[31]Yuille A.L.,Cohen D.S.,and Halliman P.W.Feature extraction from faces using deformable templates[A],In Proc.Computer Vision and Pattern Recognition[C],1999.
[32]Fu H.C.,Lai P.S.,Lou R.S.,Face detection and eye localization by neural network based color segmentation[A],In Proc.Neural Networks for Signal Processing[C],2000(2):507-516.
[33]Reinders M.J.T.,Koch R.W.C.,Gerbrands,Locating facial features in image sequences using neural networks[A],In Proc.Automatic Face and Gesture Recognition,1998.
[34]Kin C.Y.,Cipolla R.,A probabilistic framework for perceptual grouping of features for human face detection[A],In Proc.Automatic Face and Gesture Recognition[C],2000.
[35]Novak C L,Shafer S A.Supervised color constancy for machine vision.In Proceedings of SPIE,Human Vision,Vision Processing,and Digital Display,New York,1991,1453:353-368.
[2]郑庆．基于肤色的人脸检查与人眼定位[D]．电子科技大学,2007．5-6,25-26．
[3]Erik Hjelmas,Boon Kee Low.Face detection:a survey computer vision and image understanding,2001,83:236-274.
[4]M.Turk,Pentland.Eigenfaces for recognition.J.of cognitive Neuroscience,1991,3(1):71-76.
[5]N.Intrator,D.Reisfeld,Y.Yeshurun.Face recognition using a hybrid supervised unsupervised neural-network,PRL (17),January 1996,No.Ⅰ.:67-76.
[6]H.A.Rowley,S.Baluja,T.kanade.Rotation invariant neural network-based face detection.IEEE Transaction on Pattern Analysis and Machine Intelligence.1998.
[7]E.Osuna,R.Freund,F.Girosi.Training support vector machines:an application to face detection.Proc.IEEE Conf.on Computer Vision and Pattern Recognition.June 1998.
[8]N.Littleston.Learning quickly when irrelevant attributes bound:a new linear-threshold algorithm.Machine learning,1998.Vol.2:285-278.
[9]Rainer Lienhart and Jochen Maydt.An extended set of haar-like features for rapid object detection.IEEE ICIP 2002,Sep.2002.Vol.1:900-903.
[10]Paul Viola,Michael Jones.Rapid object detection using a boosted cascade of simple features.IEEE Computer Vision and Pattern Recognition (CVPRV'01).2001.Vol.l:511-518.
[11]P.Belhummer,J.Hespanha,D.Kriegman.Eigenfaces versus fisher-faces:recognition using class linear projection.IEEE Trans,on Pattern Analysis and machine Intelligence.1997.19(7):711-720.
[12]F.Samaria,S.Young.HMM based architecture for face identification.Image and Vision Computing.1994.Vol.(12):537-583.
[13]K-K Sung.Learning and example selection for object and pattern detection,PhD Thesis,Massachusetts Institute of Technology.1996.
[14]AN Rajagopalan,KS Kumar,J.Karlekar,R.Manivasakan,and MM Patil.Finding faces in photographs.In Proceedings of the Sixth International Conference on Computer Vision.1998.
[15]M.H.Yang,N.Ahuja.Detection human Faces in Color Images.In IEEE International Conference on Image Processing,Chicago,IL,October 4-7 1998:127-130.
[16]E.Saber,A.M.Tekalp.Frontal-view face detection and facial feature extraction using color,shape and symmetry based cost functions.Pattern Recognition Letters.1998.19(8): 669-680.
[17]J.Cai,A.Goshasby.Detecting human faces in color images.Image and Vision Computing.1999.18:63-74.
[18]王文宁,王汇源。彩色图像中多姿态的人脸快速检测与定位[J],计算机工程,2005,31(7):149-151.
[19]徐战武,朱森良.基于颜色的皮肤检测综述[J],中国图形图像学报,2007,3(12):203-209。
[20]彭波,崔永普,黄丹霞等.基于肤色和唇色信息的人脸检测方法的研究[J],计算机工程与设计,2006,26(6):1500-1502.
[21]陈泽宇,戚飞虎,陈刚.利用颜色信息的人脸检测方法[[J],上海交通大学学报,2000,34(6):793-795.
[22]党治,冯晓毅.基于新的肤色模型的人脸检测方法[J],计算机应用,2006,26(3):615-617
[23]Haralick R,Zhuang X.Image analysis using mathematical morphology.IEEE Trans.On Pattern Analysis and Machine Intelligence.1997,9(4):532-550.
[24]崔屹,图像处理与分析-数理形态学方法及应用[M],北京科学出版社,2000。
[25]Matheron G..Random Sets and Integral Geometry[M].New York,1995.
[26]T Martin,Hagan Howard,Mark Demuth,H.Beale,神经网络设计[M].机械工业出版社,2002,9。
[27]Tom M,Mitchell。机器学习[M],北京,机械工业出版社,2003.
[28]阮秋琦,数字图像处理学[M],电子工业出版社,2001。
[29]董长虹,赖志国,余啸海,MATLAB图像处理与应用[M],国防工业出版社,2004:182-195
[30]Feng G.C,Yuen P.C.,Variance projection function and its application to eye detection for human face recognition[J],Pattern Recognition Letters,1998,19:899-906.
[31]Yuille A.L.,Cohen D.S.,and Halliman P.W.Feature extraction from faces using deformable templates[A],In Proc.Computer Vision and Pattern Recognition[C],1999.
[32]Fu H.C.,Lai P.S.,Lou R.S.,Face detection and eye localization by neural network based color segmentation[A],In Proc.Neural Networks for Signal Processing[C],2000(2):507-516.
[33]Reinders M.J.T.,Koch R.W.C.,Gerbrands,Locating facial features in image sequences using neural networks[A],In Proc.Automatic Face and Gesture Recognition,1998.
[34]Kin C.Y.,Cipolla R.,A probabilistic framework for perceptual grouping of features for human face detection[A],In Proc.Automatic Face and Gesture Recognition[C],2000.
[35]Novak C L,Shafer S A.Supervised color constancy for machine vision.In Proceedings of SPIE,Human Vision,Vision Processing,and Digital Display,New York,1991,1453:353-368.