人脸识别技术是机器视觉和图像模式识别的一个重要研究课题，在安全领域、生物领域和侦察领域有广泛的应用前景，如身份验证、指纹虹膜识别和监控系统等。人脸识别技术主要包括人脸检测、特征提取和分类识别，其中特征提取和分类识别是研究的重点，也是提高识别率的关键。图像分类识别的本质是建立一个能够识别出图像中感兴趣物体类别的表示系统，光照变化、旋转、遮挡、复杂背景等因素直接影响分类的准确率。如何消除这些外在因素的影响，提高分类识别的准确率成为当今学者的研究方向。图像的分类识别过程包括特征提取、特征向量在特征空间内进行相似性检索识别以及利用分类器进行图像的分类识别。

       在图像的特征提取方面，基于纹理特征的提取方法和梯度方向直方图特征的提取方法已分别被广泛应用于人脸识别研究中，已有学者证明了两种方法都有各自的优越性。本文分别对两种不同的特征提取方法进行分层融合，并通过实验数据分析识别效果，在原始方法的基础上提出新的融合特征提取方法，进一步提高识别率。

       本文对纹理特征中的局部二值模式(Local Binary Pattern, LBP)、中心对称局部二值模式(Center-Symmetric Local Binary Pattern, CS-LBP)和梯度方向特征(Histogram of Oriented Gradients, HOG)作了详细的介绍，并将两种方法进行分层融合应用于人脸识别，取得了较好的效果。针对LBP和CS-LBP描述子提取的纹理特征有限且不能有效描述图像边缘和方向信息的问题，本文主要对纹理特征和梯度特征提取和融合方法进行了改进，具体工作如下：

       1．本文针对特征提取算法中的LBP和HOG算子做出了改进。由于LBP和HOG对人脸图像进行特征提取，都是对灰度图像进行运算，而单层特征并不能消除光照等影响，本文提出对两种特征分别进行分层的改进处理，分层能够提取图像更完整的特征。

       2．针对纹理特征和梯度特征都只能单独提取纹理特征和边缘结构特征的问题，本文做出了改进。由于LBP能有效的提取图像的纹理特征，HOG能准确地描述图像的边缘结构特征和方向信息，本文提出了将两种方法进行串连级联，进一步应用分层特征进行融合。

       3．对于原始人脸识别方法中利用欧氏距离进行相似性度量的问题，本文做出了改进。原始识别阶段是利用欧氏距离进行相似性度量，本文提出了利用卡方距离统计直方图之间的距离特征，并通过实验将两种方法进行对比分析。

       4．LBP和HOG描述子都是对人脸的分块区域进行运算，对于分块大小的影响和边缘无法计算到的问题，本文做出了改进。本文比较了不同分块下识别率的高低，选择最优分块，且在图像外边缘补充像素，保证边缘像素可以参与运算。

[1] Didaci L, Marcialis G L, Roli F. Analysis of unsupervised template update in biometric recognition systems [J]. Pattern Recognition Letters, 2014(37): 151-160.

[2] Yang M H, Kriegman D J, Ahuja N. Detecting face in image: A survey [J]. IEEE Transaction on Pattern Analysis and Machine Intelligence, 2002, 24(1): 34-54.

[3] Chellappa R, Wilson C L, Sirohey S. Human and machine recognition of faces:a survey[J]. Proceeding of the IEEE, 1995, 83(5): 705-740.

[4] Zhao W, Chellappa R, Phillips Jonathon P. Face recognition: A literature survey[J]. ACM Computing Surveys (CSUR), 2003, 35(4): 399-458.

[5] Chellapa R, Sinha P, Phillips Jonathon P. Face recognition by computers and humans [J]. IEEE Computer Society, 2010, 43(2): 46-55.

[6] Ricanek K, Sawides M, et al. Unconstrained biometric identification: emerging technologies [J]. IEEE Computer Society, 2010, 43(2): 56-63.

[7] 杨淑平. 人脸特征提取与识别算法研究[D].长沙：中南大学数学与统计学院，2013.

[8] Rigas I, Economou G, Fotopoulos S. Biometric identification based on the eye movements and graph matching techniques[J]. Pattern Recognition Letters, 2012, 33(6): 786-792.

[9] Nasta C, Mitschke M. Real-time face recognition using feature combination[C]. Third IEEE International Conference on Automatic Face and Gesture Recognition, 1998: 312-317.

[10] Calton F. Personal identification and description [J]. Nature, 1888: 173-177.

[11] Parke F I. Computer generated animation of faces [J]. Proceedings of the ACM annual conference, 1972, (1): 451-457.

[12] Bledsoe W. Some results on multicategory pattern recognition [J]. Journal of the ACM (JACM), 1966, 13(2): 304-316.

[13] 王海炳. 人脸识别技术研究与实现[D]. 武汉：武汉理工大学，2008.

[14] Kanade T. Picture processing system by computer complex and recognition of human faces [D]. Kyoto: Kyoto University, 1974.

[15] Sirovich L, Kirby M. Low-dimension procedure for the characterization of human faces [J]. JOSAA, 1987, 4(3): 519-524.

[16] Turk M, Pentland A. Eigenfaces for recognition [J]. Journal of Cognitive Neuroscience, 1991, 3(1): 71-86.

[17] Belhumeur P N, Hespanha J P, Kpiegman D. Eigenfaces vs. fisherfaces: Recognition using class specific linear projection [J]. IEEE Transaction on Pattern Analysis and Machine Intelligence, 1997, 19(7): 711-720.

[18] Cootes T F, Taylor C J, Cooper D H, et al. Active shape models-their training and application[J]. Computer Vision and Image Understanding, 1995, 61(1):38-59.

[19] Cootes T F, Edwards G J, Taylor C J. Active appearance models[C]//ECCV. 1998, 2: 484-498.

[20] Wiskott L, Fellous J M, Kuiger N, et al. Face recognition by elastic bunch graph matching [J]. IEEE Transaction on Pattern Analysis and Machine Intelligence, 1997, 19(7): 775-779.

[21] Brunelli R, Poggio T. Feature recognition: features versus templates [J]. IEEE Transaction on Pattern Analysis and Machine Intelligence, 1993, 15(10): 1042-1052.

[22] 柯江民. 人脸图像的监测和识别的研究与实现[D]. 成都：成都理工大学，2009.

[23] 叶慧星. 基于ASLBP的人脸识别研究[D]. 浙江：浙江大学信息与电子工程学院，2014.

[24] Wallraven C, Schwaninger A, Bulthoff H H. Learning from humans: computational modeling of face recognition [J]. Network: Computation in Neural Systems, 2005, 16(4): 401-418.

[25] Kepenekci B, Tek F B , Akar G B. Occluded face recognition based on Gabor wavelets[C]. In: Proceeding of the IEEE International Conference on Image Processing (ICIP '02), Rochester, NY, USA, 2002, 1: 293-296.

[26] Albiol A, Monzo D, Martin A, et al. Face recognition using HOG-EBGM[J]. Pattern Recognition Letters, 2008, 29(10):1537-1543.

[27] Tan X Y, Triggs B. Fusion Gabor and LBP feature sets for kernel-based face recognition [C]. //International Conference on Analysis and Modeling of Faces and Gestures. Heidelberg: Springer, 2007: 235-249.

[28] Park W J, Kim D H, Lyuh C G. Fast human detection using selective block-based HOG-LBP [C]. International Conference on Image Processing (ICIP), 2012 601-604.

[29] Cosma C, Brehar R, Nedevschi S. Pedestrians detection using a cascade of LBP and HOG classifiers[C]. International Conference on Computational Intelligence and Security (CIS), 2011, 1184-1187.

[30] 黄非非. 基于LBP的人脸识别研究[D]. 重庆：重庆大学，2009.

[31] 汤德俊. 人脸识别中图像特征提取与匹配技术研究[D]. 大连：大连海事大学，2013.

[32] 陆华. 基于局部二值模式的人脸识别和表情识别研究[D]. 山东：山东大学信息科学与工程学院，2014.

[33] 山世光,高文,陈熙霖.基于面部纹理分布和变形模板的面部特征提取[J].软件学报, 2001, 12(4): 570-577.

[34] Nikolaidis A, Pitas I. Facial feature extraction and pose determination[J]. Pattern Recognition, 2000, 33(11):1783-1791.

[35] Pang Y, Shang M, Yuan Y, et al. Scale invariant image matching using triplewise constraint and weighted voting [J], Neurocomputing ,2012,83:64-71.

[36] Ahonen T, Hadid A, Peitikainen M. Face deescription with local binary patterns: application to face recognition, IEEE Transaction on Pattern Analysis and Machine Intelligence, 2006: 2037-2041.

[37] He D C, Wang L. Texture unit, texture spectrum, and texture analysis[J].IEEE Transactions on Remote Sensing, 1990, 28(4): 509-512.

[38] Bartlett M S, Movellan J R, Sejnowski T J. Face recognition by independent component analysis[J]. IEEE Transactions on Neural Networks,2002, 13(6): 1450-1464.

[39] Etenmad K, Chellappa R. Discriminant analysis for recognition of human faceimages [J].Journal of the Optical Society of America,1997, 14(8):1724-1733.

[40] Hyvarinen A. Fast and robust fixed-point algorithms for independent component analysis[J]. IEEE Transactions on Neural Networks, 1999, 10(3):626-634.

[41] Michael K, Ming He, Garrison W. Categorization of faces using unsupervised feature extraction[C], International Joint Conference on Neural Networks. 1990, 2: 65-70.

[42] 王晓云.基于支持向量机的优化在图像分类中的应用研究[D], 上海：上海交通大学，2012.

[43] Sinha P, Balas B, Ostrovsky Y, et al. Face recognition by humans: nineteen results all computer vision researchers should know about[J]. In: Proceeding of the IEEE. 2006, 94(11): 1948-1962.

[44] Moghaddam B, Pentland A. Probabilistic visual learning for object representation [J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 1997, 19(7): 696-710.

[45] Buhmann J, Lades M, Malsburg C V D. Size and distortion invariant object recognition by hierarchical graph matching [C]. In: Proceeding of the international joint conference on neural networks, 1989:155-159.

[46] Lades M, Vorbmggen J C, Buhmann J, et al. Distortion invariant object recognition in the dynamic link architecture[J]. IEEE Transactions on Computers, 1993, 42(3):300-310.

[47] Su Y, Shan SH G, Chen X L, et al. Hierarchical ensemble of global and local classifiers for face recognition[J]. IEEE transaction on image processing, 2009, 18(8): 1885-1997.

[48] 熊承义，李丹婷，笪邦友. 基于 LBP 和 PCA 特征提取的人脸识别[J]. 中南民族大学学报(自然科学版)，2011, 30(2): 75-79.

[49] 高志升，袁红照，杨军. 融合CDI和LBP的人脸特征提取与识别算法[J]. 光电子˙激光，2010,21(1): 112-115.

[50] 卢建云. 基于Gabor小波与CS-LBP的人脸特征提取算法研究[D]. 重庆：重庆大学计算机学院，2010.

[51] 王玮，黄非非，李见为，等. 使用多尺度LBP特征描述与识别人脸[J].光学精密工程, 2008, 16(4): 696-705.

[52] Yu J, Qin Z CH, Wan T,et al. Feature integration analysis of bag-of-features model for image retrieval [J], Neurocomputing, 2013,120:355-364.

[53] Ojala T, Pietik?inen M, Harwood D. A comparative study of texture measures with classification based on feature distributions[J]. Pattern Recognition, 1996, 29(1): 51-59.

[54] Ahonen T, Hadid A, Peitikainen M. Face deescription with local binary patterns: application to face recognition, IEEE Transaction on Pattern Analysis and Machine Intelligence, 2006: 2037-2041.

[55] 毋小省.改进的旋转不变区域纹理谱描述符[J].光电子˙激光, 2011, 22(5): 783-787.

[56] 宋克臣,颜云辉,陈文辉,等.局部二值模式方法研究与展望[J].自动化学报，2013,39(6):731-744.

[57] 孙君顶，毋小省. 纹理谱描述符及其在图像检索中的应用[J].计算机辅助设计与图形学学报, 2010, 22(3): 516-520.

[58] Pietik?inen M, Ahonen T, Takala V. Block-based methods for image retrieval using local binary patterns[C]. in: Proceedings of the 14th Scandinavian Conference on Image Analysis,2005:882-891.

[59] Wang X Y, Han X, Yan S C. An HOG-LBP human detector with partial occlusion handling [C]. ICCV, 2009: 32-39.

[60] He D C, Wang L. Texture unit, texture spectrum, and texture analysis[J].IEEE Transactions on Remote Sensing, 1990, 28(4): 509-512.

[61] Heikkil? M, Pietik?inen M, Schmid C. Description of interest regions with local binary patterns[J]. Pattern Recognition, 2009,42(3):425-436.

[62] Zheng Y, Shen C, Hartley R. Effective pedestrian detection using center-symmetric local binary/trinary patterns[C],in ACCV,pp.281-292,2010.

[63] 刘丹,胡永健,刘琲贝.联合SIFT特征点和CS-LBP特征描述子的复制粘贴篡改检测[J].合肥工业大学学报(自然科学版),2012,35(3):325-330.

[64] Lowe D G. Object recognition from local scale-invariant features [C], International Conference on Computer Vision, 1999: 1150-1157.

[65] 向征，谭恒良，马争鸣. 改进的HOG和Gabor，LBP性能比较[J].计算机辅助设计与图形学学报, 2012, 24(6): 787-792.

[66] 赵帅. 基于SIFT算子的模糊图像识别算法[D]. 苏州：苏州大学，2014.

[67] Dalal N, Triggs B. Histograms of oriented gradients for human detection[C]//Proceedings of the 2005 IEEE Conference on Computer Vision and Pattern Recognition. San Diego: IEEE Computer Society Press, 2005:886-893.

[68] Déniz O, Bueno G, Salido J, et al. Face recognition using histograms of oriented gradients[J]. Pattern Recognition Letters, 2011,32(12): 1598-1603.

[69] 陈锐，彭启民. 基于稳定区域梯度方向直方图的行人检测方法[J]. 计算机辅助设计与图形学学报, 2012, 24(3): 372-377.

[70] 张剑，何骅，詹小四，等. 用SIFT词汇树实现的姿态无关的人脸识别[J]. 计算机辅助设计与图形学学报，2013, 25(11): 1694-1700.

[71] 贾伟，王正勇，张杰，等. 一种基于改进的CS-LBP算子纹理图像自适应检索方法[J].微电子学与计算机, 2013, 30(9): 75-78.

[72] 尚俊. 基于HOG特征的目标识别算法研究[D]. 武汉: 华中科技大学软件学院, 2012.

[73] 万源，李欢欢，童恒庆，吴克风. CS-LBP和HOG分层特征融合的人脸识别研究[J]. 武汉理工大学学报(交通科学与工程版), 2014, 38(4): 801-805.