基于小波变换的正则化人脸预处理和演化人脸识别
|
摘要
人脸识别研究起源于众多科技工作者追求计算机人性化的美好构想，目标是赋予计算机根据人脸自发辨别人物身份的能力，具有极其重要的科学意义和非常巨大的应用价值。人脸识别作为一个科学问题，是一个典型的图像处理、模式分析、理解与分类计算问题，涉及模式识别、自然与演化计算、计算机视觉、智能人机交互、图形学、认知科学等多个学科。作为生物信息识别关键技术之一的人脸识别技术在国家安全、公共安全、信息安全、金融等多种领域具有巨大的应用前景。
经过三十多年的发展，人脸识别技术取得了长足的进步，目前最好的人脸识别系统在理想情况下已经能够取得可以被研究人员接受的识别性能，但众多人脸识别系统理论、应用研究、开发、市场工作者和用户的测试和实践经验表明：非理想条件下(包括多种变化因素影响下)的人脸识别仍未较成熟。要开发出真正广泛应用于市场的鲁棒、实用的人脸识别应用系统还需要解决大量的理论和应用技术问题，尤其需要研究核心问题：(1)作为实际应用识别必要前提条件的人脸关键特征检测和精确定位；(2)人脸预处理和高效鲁棒的人脸特征描述识别算法。
本文重点探讨了人脸识别中的上述核心问题之一：人脸预处理和高效鲁棒的人脸特征描述识别算法。本文的主要研究工作包括如下几个方面：
1．总结了现有的主要人脸图像数据库的情况，讨论了人脸识别领域目前仍面临的重要开放问题之一：人脸特征描述。
2．推导出了人脸预处理中的亮度归一的相对完整表达式，较详细地重新推导基于小波变换的正则化方法并将其第一次应用于人脸预处理中的降质人脸恢复，实验恢复效果在一定程度上较优于传统常用正则化方法。
3．研究了Gabor小波网络人脸特征描述方法，对其进行自然演化优化，提出和实现了IOEA-GWN人脸识别新算法，实验显示在一定程度上提高了受多种变化因素影响下的人脸识别率。
4．对本文人脸识别的后续工作做了分析和展望，得出结论：可以通过例如GEP等各种新型演化的方法来提高GWN优化效果从而提高人脸识别率，可启发相关和其他模式识别领域的研究者改进或创新应用于各种模式识别领域中。
Face Recognition Research comes of the nice concept (humanization of computers of thousands of scientific workers. It aims at endowing computers with the ability to identify different human beings according to his/her face image. The research has both significant theoretic values and wide market applications. Face Recognition problem is a typical image processing, pattern analysis, understanding and classificati on problem, closely related to many disciplines such as Pattern Recognition, Natural and Evolutionary Computation, Computer Vision, Intelligent Human-Computer Interaction, Computer Graphics, and Cognitive Psychology etc. As one of key techniques of biological information recognition, Face Recognition has significant application in several areas of national security, public security, information security, finance etc. After more than 30 years' development, Face Recognition has made great progress. The state-of-the-art Face Recognition system can perform identification successfully under well-controlled environment. However, evaluation results and practictical experience have shown that Face Recognition technologies are currently far from mature. The following key issue is especially pivotal: (1) The accurate facial feature location problem, which is the prerequisite for sequent feature exaction and Classification; (2) Face Preprocessing and efficient face representation and recognition methods.
In this thesis, the second of the above-mentioned key issues are studied.
The main contribution of this thesis includes:
1. The thesis summarizes existing main face image database and discuss the important problem to be confronted with in Face Recognition area: Face characteristic Description.
2. The thesis derives the whole expression comparatively and takes the lead in using a Regularization method based on wavelet transform as an important degraded face restoring method in face preprocessing. The results show better effect than common Regularization method to some extent.
3. The thesis discusses the Gabor Wavelet Networks and deal with it by evolutionary optimization and put forward and implement new face recognition methods: IOEA-GWN. Experiments shows higher recognition rates under various changed factors.
4. The thesis gives out analysis and prospect of further studies and draw the conclusion: GWN can be optimized by several evolutionary algorithms to improve the effect of face recognition rates and these methods can be used by researchers from other pattern recognition area.
经过三十多年的发展，人脸识别技术取得了长足的进步，目前最好的人脸识别系统在理想情况下已经能够取得可以被研究人员接受的识别性能，但众多人脸识别系统理论、应用研究、开发、市场工作者和用户的测试和实践经验表明：非理想条件下(包括多种变化因素影响下)的人脸识别仍未较成熟。要开发出真正广泛应用于市场的鲁棒、实用的人脸识别应用系统还需要解决大量的理论和应用技术问题，尤其需要研究核心问题：(1)作为实际应用识别必要前提条件的人脸关键特征检测和精确定位；(2)人脸预处理和高效鲁棒的人脸特征描述识别算法。
本文重点探讨了人脸识别中的上述核心问题之一：人脸预处理和高效鲁棒的人脸特征描述识别算法。本文的主要研究工作包括如下几个方面：
1．总结了现有的主要人脸图像数据库的情况，讨论了人脸识别领域目前仍面临的重要开放问题之一：人脸特征描述。
2．推导出了人脸预处理中的亮度归一的相对完整表达式，较详细地重新推导基于小波变换的正则化方法并将其第一次应用于人脸预处理中的降质人脸恢复，实验恢复效果在一定程度上较优于传统常用正则化方法。
3．研究了Gabor小波网络人脸特征描述方法，对其进行自然演化优化，提出和实现了IOEA-GWN人脸识别新算法，实验显示在一定程度上提高了受多种变化因素影响下的人脸识别率。
4．对本文人脸识别的后续工作做了分析和展望，得出结论：可以通过例如GEP等各种新型演化的方法来提高GWN优化效果从而提高人脸识别率，可启发相关和其他模式识别领域的研究者改进或创新应用于各种模式识别领域中。
Face Recognition Research comes of the nice concept (humanization of computers of thousands of scientific workers. It aims at endowing computers with the ability to identify different human beings according to his/her face image. The research has both significant theoretic values and wide market applications. Face Recognition problem is a typical image processing, pattern analysis, understanding and classificati on problem, closely related to many disciplines such as Pattern Recognition, Natural and Evolutionary Computation, Computer Vision, Intelligent Human-Computer Interaction, Computer Graphics, and Cognitive Psychology etc. As one of key techniques of biological information recognition, Face Recognition has significant application in several areas of national security, public security, information security, finance etc. After more than 30 years' development, Face Recognition has made great progress. The state-of-the-art Face Recognition system can perform identification successfully under well-controlled environment. However, evaluation results and practictical experience have shown that Face Recognition technologies are currently far from mature. The following key issue is especially pivotal: (1) The accurate facial feature location problem, which is the prerequisite for sequent feature exaction and Classification; (2) Face Preprocessing and efficient face representation and recognition methods.
In this thesis, the second of the above-mentioned key issues are studied.
The main contribution of this thesis includes:
1. The thesis summarizes existing main face image database and discuss the important problem to be confronted with in Face Recognition area: Face characteristic Description.
2. The thesis derives the whole expression comparatively and takes the lead in using a Regularization method based on wavelet transform as an important degraded face restoring method in face preprocessing. The results show better effect than common Regularization method to some extent.
3. The thesis discusses the Gabor Wavelet Networks and deal with it by evolutionary optimization and put forward and implement new face recognition methods: IOEA-GWN. Experiments shows higher recognition rates under various changed factors.
4. The thesis gives out analysis and prospect of further studies and draw the conclusion: GWN can be optimized by several evolutionary algorithms to improve the effect of face recognition rates and these methods can be used by researchers from other pattern recognition area.
引文
[1] 山世光．人脸识别中若干关键问题的研究：[博士学位论文]．北京：中国科学院研究生院，2004
[2] Chan H., Bledsoe W. A man-machine facial recognition system: some preliminary results, Technical report, Panoramic Research Inc., Cal, 1965
[3] Phillips P.J., Moon H., Rizvi S.A., et al. The FERET Evaluation Methodology for Face-Recognition Algorithms. IEEE Transaction on Pattern Analysis and Machine Intelligence, 2000. 1090～1104
[4] Penev P., Atick J. Local Feature Analysis: A General Statistical Theory for Object Representation. Network: Computation in Neural Systems, 1996. 7:477～500
[5] 覃志祥．人脸检测与识别技术研究：[硕士学位论文]．武汉：武汉大学，2005
[6] Georghiades A.S., Kriegman D.J. , Belhumeur P.N. Illumination Cones For Recognition Under Variable Lighting: Faces. Proceedings of IEEE CVPR, 1998. 52～58
[7] Georghiades A.S., Belhumeur P.N., Kriegman D.J. From few to many: Generative models for recognition under variable pose and illumination. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2001. 23(6): 643～660
[8] Matthew Turk, Alex Pentland. Eigenfaces for Recognition. Journal of Cognitive Neuroscience, 1991. 3(1): 71～86
[9] Belhumeur P.N., Hespanha J.P., Kriegman D.J. Eigenfaces vs. Fisherfaces: Recognition Using Class Specific Linear Projection. IEEE Transactions on Pattern Analysis and Machine Intelligence, 1997. 19(7): 711～720
[10] Terence Sim, Simon Baker, Maan Bsat. The CMUPose, Illumination and Expression Database. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2003, 25(12): 1615～1618
[11] Samaria F.S., Harter A.C. Parameterization of a Stochastic Model for Human Face Identification. In: Proceedings of the 2nd IEEE Workshop on Applications of Computer Vision, Sarasoto, FL, 1994
[12] Bailly-Bailliere E., Bengio S., Bimbot F., et al. The BANCA Database and Evaluation Protocol. In: Fourth International Conference on Audio- and Video-based Biometric Person Authentication (AVBPA),Guildford, UK, 2003. 625～638
[13] Hwang B. W.,Byun H.,Roh M.C.,Lee S. W. Performance Evaluation of Face Recognition Algorithms on the Asian Face Database, KFDB. In: Fourth International Conference on Audio- and Video-based Biometric Person Authentication (AVBPA), Guildford, UK, 2003. 557～565
[14] Blanz V., Vetter T. A Morphable Model for the Synthesis of 3D Faces. In: Computer Graphics Proceedings, Annual Conference Series (SIGGRAPH), Los Angeles, USA, 1999. 187～194
[15] Messer K., Matas J., Kittler J., et al. XM2VTSDB: The Extended M2VTS Database. In: Second International Conference on Audio and Video-based Biometric Person Authentication, Washington DC, USA, 1999. 72～77
[16] 崔锦泰．小波分析导论，程正兴译．西安：西安交通大学出版社，1997．2～31
[17] 王斌．应用基因算法于基于面部整体特征的人脸识别技术研究：[博士学位论文]．北京：北京邮电大学，2004
[18] Ingrid Daubechies．小波十讲，李建平等译．北京：国防工业出版社，2004．9～10
[19] 曾三友，康立山，丁立新等．一种基于正则化方法的准最佳图象复原技术．软件学报，2003．14(3)：689～696
[20] Katsaggelos A.K. Digital Image Restoration. Berlin: Springer-Verlag, 1991
[21] Andrews H., Hunt B. Digital Image Restoration. Englewood Cliff, NJ: Prentice-Hall, 1977
[22] 黄利．基于GEP的偏微分方程反问题参数识别：[硕士学位论文]．武汉：武汉理工大学，2006
[23] Berter M., Poggio T., Torre V. Ill-Posed problems in early vision. In: Proceedings of the IEEE, 1988. 76(8): 869～889
[24] Calvetti D., Reichel L., Sgallari F., et al. A regularizing Lanczos iteration method for undetermined linear systems. Journal of Computation and Applied Mathematics, 2000. 115:101～120
[25] Hansen P.C., O' Leafy D.P. The use of the L-curve in the regularization of discrete ill posed problems. SIAM Journal on Scientific Computing, 1993. 14: 1487～1503
[26] Tekalp A., Kaufman H., Woods J. Edge-Adaptive Kalman filtering for image restoration with ringing suppression. IEEE Transactions on Acoustic Speech and Signal Processing, 1989. 37(6): 892～899
[27] Qian W., Clarke L.P. Wavelet-Based neural network with fuzzy-logic adaptivity for nuclear image restoration. In: Proceedings of the IEEE, 1996. 84(10): 1458～1473
[28] Wong H.S., Ling G. Application of evolutionary programming to adaptive regularization in image restoration. IEEE Transactions on Evolutionary Computation, 2000. 4(4): 309～326
[29] 管致中，夏恭珞．信号与线性系统(第三版)下册．北京：高等教育出版社，2002．272～274
[30] Bertero M., Boccacci P. Introduction to Inverse Problems in Imaging. Dirac House, Temple Back, Bristol BSI 6BE, UK: Institute of Physics Publishing, 1998
[31] Volker Krueger, Gerald Sommer. Gabor Wavelet Networks for Object Representation. Journal of the Optical Society of America (JOSA), 2002. 19(6): 1112～1119
[32] Cootes T.F., Edwards G.J., Taylor C.J. Active appearance models. In: Proceedings European Conference on Computer Vision. Germany: Freiburg, 1998. 2:484～498
[33] Calderon A.P. Intermediate spaces and interpolation, the complex method. Stud. Math., 1964. 24:113～190
[34] Louis A., Maas P., Riedler A. Wavelet: Theorie und Anwendung. Teubner, 1994
[35] Ingrid Daubechies. Orthonormal bases of compactly supported wavelets. Communications on Pure and Applied Mathematics, 1988. 41(7): 909～996
[36] Lee T.S. Image representation using 2D Gabor wavelets. IEEE Transaction on Pattern Analysis and Machine Intelligence, 1996. 18:959～971
[37] Gabor D. Theory of Communications. Journal of Inst of Electronical Engineering, 1946. 93:429～457
[38] Wiskott L., Fellous J.M., Kruger N., et al. Face recognition by elastic bunch graph matching. IEEE Transaction on Pattern Analysis and Machine Intelligence, 1997. 19: 775～779
[39] Candida Ferreira. Gene Expression Programming: A New Adaptive Algorithm for Solving Problems. Complex Systems, 2001. 13(2): 87～129
[40] 康卓，李艳，黄竞伟等．程序重用的自动程序设计方法．武汉大学学报(理学版)，2006．52(5)：649～654
[41] 詹炜，戴光明，龚文引．求解函数优化问题的一种高效混合演化算法．计算机工程与应用，2006．70～72
[2] Chan H., Bledsoe W. A man-machine facial recognition system: some preliminary results, Technical report, Panoramic Research Inc., Cal, 1965
[3] Phillips P.J., Moon H., Rizvi S.A., et al. The FERET Evaluation Methodology for Face-Recognition Algorithms. IEEE Transaction on Pattern Analysis and Machine Intelligence, 2000. 1090～1104
[4] Penev P., Atick J. Local Feature Analysis: A General Statistical Theory for Object Representation. Network: Computation in Neural Systems, 1996. 7:477～500
[5] 覃志祥．人脸检测与识别技术研究：[硕士学位论文]．武汉：武汉大学，2005
[6] Georghiades A.S., Kriegman D.J. , Belhumeur P.N. Illumination Cones For Recognition Under Variable Lighting: Faces. Proceedings of IEEE CVPR, 1998. 52～58
[7] Georghiades A.S., Belhumeur P.N., Kriegman D.J. From few to many: Generative models for recognition under variable pose and illumination. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2001. 23(6): 643～660
[8] Matthew Turk, Alex Pentland. Eigenfaces for Recognition. Journal of Cognitive Neuroscience, 1991. 3(1): 71～86
[9] Belhumeur P.N., Hespanha J.P., Kriegman D.J. Eigenfaces vs. Fisherfaces: Recognition Using Class Specific Linear Projection. IEEE Transactions on Pattern Analysis and Machine Intelligence, 1997. 19(7): 711～720
[10] Terence Sim, Simon Baker, Maan Bsat. The CMUPose, Illumination and Expression Database. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2003, 25(12): 1615～1618
[11] Samaria F.S., Harter A.C. Parameterization of a Stochastic Model for Human Face Identification. In: Proceedings of the 2nd IEEE Workshop on Applications of Computer Vision, Sarasoto, FL, 1994
[12] Bailly-Bailliere E., Bengio S., Bimbot F., et al. The BANCA Database and Evaluation Protocol. In: Fourth International Conference on Audio- and Video-based Biometric Person Authentication (AVBPA),Guildford, UK, 2003. 625～638
[13] Hwang B. W.,Byun H.,Roh M.C.,Lee S. W. Performance Evaluation of Face Recognition Algorithms on the Asian Face Database, KFDB. In: Fourth International Conference on Audio- and Video-based Biometric Person Authentication (AVBPA), Guildford, UK, 2003. 557～565
[14] Blanz V., Vetter T. A Morphable Model for the Synthesis of 3D Faces. In: Computer Graphics Proceedings, Annual Conference Series (SIGGRAPH), Los Angeles, USA, 1999. 187～194
[15] Messer K., Matas J., Kittler J., et al. XM2VTSDB: The Extended M2VTS Database. In: Second International Conference on Audio and Video-based Biometric Person Authentication, Washington DC, USA, 1999. 72～77
[16] 崔锦泰．小波分析导论，程正兴译．西安：西安交通大学出版社，1997．2～31
[17] 王斌．应用基因算法于基于面部整体特征的人脸识别技术研究：[博士学位论文]．北京：北京邮电大学，2004
[18] Ingrid Daubechies．小波十讲，李建平等译．北京：国防工业出版社，2004．9～10
[19] 曾三友，康立山，丁立新等．一种基于正则化方法的准最佳图象复原技术．软件学报，2003．14(3)：689～696
[20] Katsaggelos A.K. Digital Image Restoration. Berlin: Springer-Verlag, 1991
[21] Andrews H., Hunt B. Digital Image Restoration. Englewood Cliff, NJ: Prentice-Hall, 1977
[22] 黄利．基于GEP的偏微分方程反问题参数识别：[硕士学位论文]．武汉：武汉理工大学，2006
[23] Berter M., Poggio T., Torre V. Ill-Posed problems in early vision. In: Proceedings of the IEEE, 1988. 76(8): 869～889
[24] Calvetti D., Reichel L., Sgallari F., et al. A regularizing Lanczos iteration method for undetermined linear systems. Journal of Computation and Applied Mathematics, 2000. 115:101～120
[25] Hansen P.C., O' Leafy D.P. The use of the L-curve in the regularization of discrete ill posed problems. SIAM Journal on Scientific Computing, 1993. 14: 1487～1503
[26] Tekalp A., Kaufman H., Woods J. Edge-Adaptive Kalman filtering for image restoration with ringing suppression. IEEE Transactions on Acoustic Speech and Signal Processing, 1989. 37(6): 892～899
[27] Qian W., Clarke L.P. Wavelet-Based neural network with fuzzy-logic adaptivity for nuclear image restoration. In: Proceedings of the IEEE, 1996. 84(10): 1458～1473
[28] Wong H.S., Ling G. Application of evolutionary programming to adaptive regularization in image restoration. IEEE Transactions on Evolutionary Computation, 2000. 4(4): 309～326
[29] 管致中，夏恭珞．信号与线性系统(第三版)下册．北京：高等教育出版社，2002．272～274
[30] Bertero M., Boccacci P. Introduction to Inverse Problems in Imaging. Dirac House, Temple Back, Bristol BSI 6BE, UK: Institute of Physics Publishing, 1998
[31] Volker Krueger, Gerald Sommer. Gabor Wavelet Networks for Object Representation. Journal of the Optical Society of America (JOSA), 2002. 19(6): 1112～1119
[32] Cootes T.F., Edwards G.J., Taylor C.J. Active appearance models. In: Proceedings European Conference on Computer Vision. Germany: Freiburg, 1998. 2:484～498
[33] Calderon A.P. Intermediate spaces and interpolation, the complex method. Stud. Math., 1964. 24:113～190
[34] Louis A., Maas P., Riedler A. Wavelet: Theorie und Anwendung. Teubner, 1994
[35] Ingrid Daubechies. Orthonormal bases of compactly supported wavelets. Communications on Pure and Applied Mathematics, 1988. 41(7): 909～996
[36] Lee T.S. Image representation using 2D Gabor wavelets. IEEE Transaction on Pattern Analysis and Machine Intelligence, 1996. 18:959～971
[37] Gabor D. Theory of Communications. Journal of Inst of Electronical Engineering, 1946. 93:429～457
[38] Wiskott L., Fellous J.M., Kruger N., et al. Face recognition by elastic bunch graph matching. IEEE Transaction on Pattern Analysis and Machine Intelligence, 1997. 19: 775～779
[39] Candida Ferreira. Gene Expression Programming: A New Adaptive Algorithm for Solving Problems. Complex Systems, 2001. 13(2): 87～129
[40] 康卓，李艳，黄竞伟等．程序重用的自动程序设计方法．武汉大学学报(理学版)，2006．52(5)：649～654
[41] 詹炜，戴光明，龚文引．求解函数优化问题的一种高效混合演化算法．计算机工程与应用，2006．70～72