基于CURVELET变换的人脸检测
|
摘要
目前,关于人脸图像的处理包含人脸识别、人脸跟踪、姿态估计、表情识别等内容,而人脸检测是上述内容实现的基础。人脸检测问题也是计算机视觉和模式识别领域的一个研究热点。
Curvelet变换是近年来提出的多尺度变换中的一种,该变换具有很强的方向性,且能用极少的非零系数精确表达图像的边缘,对物体边缘信息具有最优稀疏表示,可以以较少的系数表达图像的特征。第一代Curvelet变换的数字实现算法冗余度较高,现在,第二代Curvelet变换的快速算法已经实现,变换速度的提高,为Curvelet变换的应用提供了有利条件。
本文针对人脸检测中的若干问题,提出了先对图像进行基于肤色分割的人脸区域预分割,然后再利用基于Curvelet变换提取的人脸特征对备选肤色区域进行验证的人脸检测方法。首先,选择YCbCr彩色空间来进行肤色区域分割,以快速去除复杂的背景,提高人脸检测的效率。其次,利用Curvelet变换提取出人脸样本的特征,得到特征脸。最后,在分割出肤色区域和提取出特征脸的条件下,对每一个疑似人脸的肤色区域进行Curvelet变换,将最小尺度的变换系数和特征脸的系数进行比较,在均方误差不超过某一阈值的情况下判断为人脸。
实验证明,本文提出的人脸检测方法是合理的,具有一定的实用价值。
Research efforts in face processing include face recognition, face tracking, pose estimation, and expression recognition, face detection and location is involved in all of these researching. Face detection has been a research focus of computer vision and pattern recognition.
Curvelet transform is one of the multiscale transforms which are implemented in recent years. It contains more directional information, can represent the edge use less coefficients, it has optimally sparse representation of objects with edges, in other words this transform can represent the eigenvalue of the image. The digital algorithm of the first generation curvelet transform is redundant. Now, the digital algorithm of the second generation curvelet transform, namely, fast discrete curvelet transform, has been implemented.
A face detector based on skin color segmenting preprocessing and curvelet transform is designed in this thesis. First, we designed a skin color segmenting approach based on YCbCr color space. Applied this approach to the preprocessing of the face detection system, it can discard background regions of the image quickly, so enhance the executing efficiency and detecting performance of the face detection system. Secondly, face feature extraction algorithm based on curvelet transform is presented, so we can represent eigenface with the coefficients of curvelet transform. Finally, take curvelet transform on every face color region which is preprocessed, then get the result of face detection through comparing the coefficients of coarse scale with the eigenface,
The results of experiments show that the detection methods proposed in this paper have excellent performances and high practicality.
Curvelet变换是近年来提出的多尺度变换中的一种,该变换具有很强的方向性,且能用极少的非零系数精确表达图像的边缘,对物体边缘信息具有最优稀疏表示,可以以较少的系数表达图像的特征。第一代Curvelet变换的数字实现算法冗余度较高,现在,第二代Curvelet变换的快速算法已经实现,变换速度的提高,为Curvelet变换的应用提供了有利条件。
本文针对人脸检测中的若干问题,提出了先对图像进行基于肤色分割的人脸区域预分割,然后再利用基于Curvelet变换提取的人脸特征对备选肤色区域进行验证的人脸检测方法。首先,选择YCbCr彩色空间来进行肤色区域分割,以快速去除复杂的背景,提高人脸检测的效率。其次,利用Curvelet变换提取出人脸样本的特征,得到特征脸。最后,在分割出肤色区域和提取出特征脸的条件下,对每一个疑似人脸的肤色区域进行Curvelet变换,将最小尺度的变换系数和特征脸的系数进行比较,在均方误差不超过某一阈值的情况下判断为人脸。
实验证明,本文提出的人脸检测方法是合理的,具有一定的实用价值。
Research efforts in face processing include face recognition, face tracking, pose estimation, and expression recognition, face detection and location is involved in all of these researching. Face detection has been a research focus of computer vision and pattern recognition.
Curvelet transform is one of the multiscale transforms which are implemented in recent years. It contains more directional information, can represent the edge use less coefficients, it has optimally sparse representation of objects with edges, in other words this transform can represent the eigenvalue of the image. The digital algorithm of the first generation curvelet transform is redundant. Now, the digital algorithm of the second generation curvelet transform, namely, fast discrete curvelet transform, has been implemented.
A face detector based on skin color segmenting preprocessing and curvelet transform is designed in this thesis. First, we designed a skin color segmenting approach based on YCbCr color space. Applied this approach to the preprocessing of the face detection system, it can discard background regions of the image quickly, so enhance the executing efficiency and detecting performance of the face detection system. Secondly, face feature extraction algorithm based on curvelet transform is presented, so we can represent eigenface with the coefficients of curvelet transform. Finally, take curvelet transform on every face color region which is preprocessed, then get the result of face detection through comparing the coefficients of coarse scale with the eigenface,
The results of experiments show that the detection methods proposed in this paper have excellent performances and high practicality.
引文
1 Yang M-H, Ahuja N, Kriegman D. A survey on face detection methods [J]. IEEE Trans.PAMI.2002, 24(1): 34-58.
2 Prag Sharma, Richard B. Reilly. a colour face image database or benchmarking of automatic face detection algorithms. EURASIP Conference focused on Video I Image Processing and Multimedia Communications. 2003, 2 (5): 423-428.
3 Lin-Lin Huang, Akinobu Shimizu, and Hidefumi Kobatake. a multi-expert approach for robust face detection. IEEE Proceedings of the 17th International Conference on Pattern Recognition .2004: 1051-1054.
4 .M.J. Seow, R. Gottumukkal, D. Valaparla, and K.V. Asari.a roubust face recognition system for real time surveillance. IEEE Proceedings of the International Conference on Information Technology: Coding and Computing .2004.
5 Jure Kovac,Peter Peer,Franc Solina.2D VERSUS 3D COLOUR SPACE FACE DETECTION.EC-VIP-MC 2003,4th EURASIP Conference focused on Video I Image Processing and Multimedia Communications. 2003.
6 Ben-Zion Shaick and Leonid Yaroslavsky. accelerating face detection by means of image segmentation.EC-VIP-MC 2003, 4th EURASIP Conference focused on Video / Image Processing and Multimedia Communications. 2003.
7 Jin Seo and Hanseok Ko. face detection using support vector domain description in color images. IEEE.2004.
8 Bernhard Froba Andreas Ernst..face detection with the modified census transform.IEEE Proceedings of the Sixth IEEE International Conference on Automatic Face and Gesture Recognition. 2004.
9 Rein-Lien Hsu, Mohamed Abdel-Mottaleb, and Anil K. Jain. FACE DETECTION IN COLOR IMAGES.
10 Bo WU, Haizhou AI, Chang HUANG and Shihong LAO.Fast Rotation Invariant Multi-View Face Detection Based on Real Adaboost. .IEEE Proceedings of the Sixth IEEE International Conference on Automatic Face and Gesture Recognition. 2004.
11 Jure Kovae , Peter Peer , Franc Solina.Illumination Independent Color-Based Face Detection. Proceedings of the 3rd International Symposium on Image and Signal Processing and Analysis.2003.
12 Marc Lievin and Franck Luthon.Nonlinear Color Space and Spatiotemporal MRF for Hierarchical Segmentation of Face Features in Video.IEEE TRANSACTIONS ON IMAGE PROCESSING. 2004, 13 (1) .
13 Evangelos Loutas, Ioannis Pitas, and Christophoros Nikou.Probabilistic Multiple Face Detection and Tracking Using Entropy Measures. IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY. 2004, 14 (1) .
14 Jie Chen, Xilin Chen, Wen Gao.Resampling for Face Detection by Self-Adaptive Genetic Algorithm. IEEE Proceedings of the 17th International Conference on Pattern Recognition . 2004.
15 Oliver Jesorsky, Klaus J. Kirchberg, and Robert W. Frischholz.Robust Face Detection Using the Hausdorff Distance. In Proc. Third International Conference on Audio- and Video-based Biometric Person Authentication, Springer, Lecture Notes in Computer Science, LNCS-2091, Halmstad, Sweden. 2001: 90-95.
16 Rong Xiao, Ming-Jing Li, and Hong-Jiang Zhang.Robust Multipose Face Detection in Images.IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY. 2004, 14 (1) .
17 Kohsia S. Huang and Mohan M. Trivedi.Robust Real-Time Detection, Tracking, and Pose Estimation of Faces in Video Streams. IEEE Proceedings of the 17th International Conference on Pattern Recognition , 2004.
18 E. J. Cand'es , D.L.Donoho. Curvelets-a surprisingly effective nonadaptive representation for objects with edges. Vanderbilt University Press. 1999: 105-120.
19 D. L. Donoho , M. R. Duncan. Digital Curvelet Transform: Strategy, Implementation, Experiments. Technical Report, Stanford University, 1999.
20 J.L.Starck, E.J.Candes, D.L.Donoho.The curvelet transform for image denoising.IEEE Trans.Im.Proc.2002;670-684.
21J.L.Starck,Fionn Murtagh,E.J.Candes,and D.L.Donoho.Gray and Color Image Contrast Enhancement by the Curvelet Transform.IEEE Trans.Im.Proc.2003;706-717.
22E.J.Candes,Laurent Demanet,D.L.Donoho and Lexing Ying.Fast Discrete Curvelet Transforms.http;//www.acm.caltech.edu/～emmanuel/publications.html.2005;1-44.
23E.J.Cand'es,F.Guo.New multiscale transforms,minimum total variation synthesis;application to edge-preserving image reconstruction.Sig.Process.2002;1519-1543.
24E.J.Candes,D.L.Donoho.Continuous Curvelet Transform Ⅱ Discretization and Frames.http;//www.acm.caltech.edu/～emmanuel/publications.html.2003;1-22.
25E.J.Candes,D.L.Donoho.Continuous Curvelet Transform;Ⅰ Resolution of the Wavefront Set.http;//www.acm.caltech.edu/～emmanuel/ publications,html.2003;1-29.
26J.-L.Starck,M.K.Nguyen and F.Murtagh,Wavelets and Curvelets for Image Deconvolution a Combined Approach,Signal Processing.2003.
27E.J.Candes,L.Demanet,The Curvelet Representation of Wave Propagators is Optimally Sparse.http;//curvelet.org/.2004.
28E.J.Candes,L.Demanet,Curvelets and Fourier Integral Operators.http;//curvelet.org/.2002.
29L.Ying,L.Demanet,E.J.Candes,3D Discrete Curvelet Transform.http;//curvelet.org/.2005.
30J.-L.Starck,E.Candes,and D.L.Donoho,Astronomical Image Representation by the Curvelet Tansform,Astronomy and Astrophysics.2003;398,785-800.
31J.L.Starck,D.L.Donoho and E.Candes,Very High Quality Image Restoration.http;//curvelet.org/.2001.
32隆刚,肖磊,陈学佺.Curvelet变换在图像处理中的应用综述.计算机研究与发展.2005年,42(8);1331-1337.
33李晖晖,郭雷,刘航.基于二代curvelet变换的图像融合研究.光学学报,2006年,第26卷(第5期).
34张强,郭宝龙.应用第二代Curvelet变换的遥感图像融合.光学精密工 程2007年,第15卷(第7期).
35金海燕,焦李成,刘芳.基于Curvelet域隐马尔可夫树模型的SAR图像去噪.计算机学报,2007年,第30卷(第3期).
36杨俊,赵忠明.基于Curvelet变换的多聚焦图像融合方法.光电工程.2007年,第34卷(第6期).
37焦李成,谭山,刘芳.脊波理论;从脊波变换到Curvelet变换[J].工程数学学报,2005年,第22卷(第5期).
38杨友生,王耀明,郭世杰.基于小波变换和支持向量机的人脸检测.微机发展.2004年,第14卷(第11期).
39李士进,朱跃龙,王志坚.基于多分类器组合的多角度彩色人脸图像检测.小型微型计算机系统,2004年,第25卷(第8期).
40李军,陈光梦.小波域中基于模糊的图像去噪方法.复旦学报(自然科学版),2005年,第44卷(第6期).
41秦树春,张淑玲,邹复民.基于运动检测和小波分析的视频图像人脸检测.科学技术与工程.2005年,第5卷(第16期).
42龙云利,刘安芝,刘希顺.基于小波变换与支持向量机的虹膜识别新算法.微计算机信息.2005年,第21卷(第12-2期).
43张强,郭宝龙.基于Curvelet变换的图像融合算法.吉林大学学报.2007,第37卷(第2期).
44肖小奎,黎绍发.加强边缘保护的Curvelet图像去噪方法[J].通信学报,2004,(2).
45曹德平.基于支持向量机的彩色人脸检测技术.华侨大学硕士学位论文.2005;40-45.
46杨清夙,游志胜,张先玉.基于豪斯多夫距离的快速多人脸检测算法.电子科技大学学报,2004年04期.
47曹刚,游志胜,刘直芳.基于小波隐性马尔可夫模型的人脸检测.信号处理.2004年,第20卷(第1期).
48倪林,Y.Miao.一种更适合图像处理的多尺度变换——Curvelet变换[J].计算机工程与应用,2004,(28).
49薛明东,郭立,张国宣,刘士建.一种新的图像特征提取算法[J].计算机应用,2004,(S1).
50邓承志,汪胜前,钟华,刘祝华,邹道文.基于Contourlet变换的图像去噪算法[J].电视技术,2004,(10).
51张忠波,马驷良,马捷.小波和神经网络在人脸光照校正中的应用.吉林大学学报(理学版)2005年,第43卷(第2期).
52陈健,周利莉,史红刚,苏大伟.一种基于Haar小波变换的彩色图像人脸检测方法.中文核心期刊《微计算机信息》(测控自动化).2005年第21卷(第10-1期).
53王虹,王秀锋.小波包分析在头肩序列的人脸检测中的应用研究.武汉理工大学学报(交通科学与工程版).2005年,第29卷(第6期).
2 Prag Sharma, Richard B. Reilly. a colour face image database or benchmarking of automatic face detection algorithms. EURASIP Conference focused on Video I Image Processing and Multimedia Communications. 2003, 2 (5): 423-428.
3 Lin-Lin Huang, Akinobu Shimizu, and Hidefumi Kobatake. a multi-expert approach for robust face detection. IEEE Proceedings of the 17th International Conference on Pattern Recognition .2004: 1051-1054.
4 .M.J. Seow, R. Gottumukkal, D. Valaparla, and K.V. Asari.a roubust face recognition system for real time surveillance. IEEE Proceedings of the International Conference on Information Technology: Coding and Computing .2004.
5 Jure Kovac,Peter Peer,Franc Solina.2D VERSUS 3D COLOUR SPACE FACE DETECTION.EC-VIP-MC 2003,4th EURASIP Conference focused on Video I Image Processing and Multimedia Communications. 2003.
6 Ben-Zion Shaick and Leonid Yaroslavsky. accelerating face detection by means of image segmentation.EC-VIP-MC 2003, 4th EURASIP Conference focused on Video / Image Processing and Multimedia Communications. 2003.
7 Jin Seo and Hanseok Ko. face detection using support vector domain description in color images. IEEE.2004.
8 Bernhard Froba Andreas Ernst..face detection with the modified census transform.IEEE Proceedings of the Sixth IEEE International Conference on Automatic Face and Gesture Recognition. 2004.
9 Rein-Lien Hsu, Mohamed Abdel-Mottaleb, and Anil K. Jain. FACE DETECTION IN COLOR IMAGES.
10 Bo WU, Haizhou AI, Chang HUANG and Shihong LAO.Fast Rotation Invariant Multi-View Face Detection Based on Real Adaboost. .IEEE Proceedings of the Sixth IEEE International Conference on Automatic Face and Gesture Recognition. 2004.
11 Jure Kovae , Peter Peer , Franc Solina.Illumination Independent Color-Based Face Detection. Proceedings of the 3rd International Symposium on Image and Signal Processing and Analysis.2003.
12 Marc Lievin and Franck Luthon.Nonlinear Color Space and Spatiotemporal MRF for Hierarchical Segmentation of Face Features in Video.IEEE TRANSACTIONS ON IMAGE PROCESSING. 2004, 13 (1) .
13 Evangelos Loutas, Ioannis Pitas, and Christophoros Nikou.Probabilistic Multiple Face Detection and Tracking Using Entropy Measures. IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY. 2004, 14 (1) .
14 Jie Chen, Xilin Chen, Wen Gao.Resampling for Face Detection by Self-Adaptive Genetic Algorithm. IEEE Proceedings of the 17th International Conference on Pattern Recognition . 2004.
15 Oliver Jesorsky, Klaus J. Kirchberg, and Robert W. Frischholz.Robust Face Detection Using the Hausdorff Distance. In Proc. Third International Conference on Audio- and Video-based Biometric Person Authentication, Springer, Lecture Notes in Computer Science, LNCS-2091, Halmstad, Sweden. 2001: 90-95.
16 Rong Xiao, Ming-Jing Li, and Hong-Jiang Zhang.Robust Multipose Face Detection in Images.IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY. 2004, 14 (1) .
17 Kohsia S. Huang and Mohan M. Trivedi.Robust Real-Time Detection, Tracking, and Pose Estimation of Faces in Video Streams. IEEE Proceedings of the 17th International Conference on Pattern Recognition , 2004.
18 E. J. Cand'es , D.L.Donoho. Curvelets-a surprisingly effective nonadaptive representation for objects with edges. Vanderbilt University Press. 1999: 105-120.
19 D. L. Donoho , M. R. Duncan. Digital Curvelet Transform: Strategy, Implementation, Experiments. Technical Report, Stanford University, 1999.
20 J.L.Starck, E.J.Candes, D.L.Donoho.The curvelet transform for image denoising.IEEE Trans.Im.Proc.2002;670-684.
21J.L.Starck,Fionn Murtagh,E.J.Candes,and D.L.Donoho.Gray and Color Image Contrast Enhancement by the Curvelet Transform.IEEE Trans.Im.Proc.2003;706-717.
22E.J.Candes,Laurent Demanet,D.L.Donoho and Lexing Ying.Fast Discrete Curvelet Transforms.http;//www.acm.caltech.edu/～emmanuel/publications.html.2005;1-44.
23E.J.Cand'es,F.Guo.New multiscale transforms,minimum total variation synthesis;application to edge-preserving image reconstruction.Sig.Process.2002;1519-1543.
24E.J.Candes,D.L.Donoho.Continuous Curvelet Transform Ⅱ Discretization and Frames.http;//www.acm.caltech.edu/～emmanuel/publications.html.2003;1-22.
25E.J.Candes,D.L.Donoho.Continuous Curvelet Transform;Ⅰ Resolution of the Wavefront Set.http;//www.acm.caltech.edu/～emmanuel/ publications,html.2003;1-29.
26J.-L.Starck,M.K.Nguyen and F.Murtagh,Wavelets and Curvelets for Image Deconvolution a Combined Approach,Signal Processing.2003.
27E.J.Candes,L.Demanet,The Curvelet Representation of Wave Propagators is Optimally Sparse.http;//curvelet.org/.2004.
28E.J.Candes,L.Demanet,Curvelets and Fourier Integral Operators.http;//curvelet.org/.2002.
29L.Ying,L.Demanet,E.J.Candes,3D Discrete Curvelet Transform.http;//curvelet.org/.2005.
30J.-L.Starck,E.Candes,and D.L.Donoho,Astronomical Image Representation by the Curvelet Tansform,Astronomy and Astrophysics.2003;398,785-800.
31J.L.Starck,D.L.Donoho and E.Candes,Very High Quality Image Restoration.http;//curvelet.org/.2001.
32隆刚,肖磊,陈学佺.Curvelet变换在图像处理中的应用综述.计算机研究与发展.2005年,42(8);1331-1337.
33李晖晖,郭雷,刘航.基于二代curvelet变换的图像融合研究.光学学报,2006年,第26卷(第5期).
34张强,郭宝龙.应用第二代Curvelet变换的遥感图像融合.光学精密工 程2007年,第15卷(第7期).
35金海燕,焦李成,刘芳.基于Curvelet域隐马尔可夫树模型的SAR图像去噪.计算机学报,2007年,第30卷(第3期).
36杨俊,赵忠明.基于Curvelet变换的多聚焦图像融合方法.光电工程.2007年,第34卷(第6期).
37焦李成,谭山,刘芳.脊波理论;从脊波变换到Curvelet变换[J].工程数学学报,2005年,第22卷(第5期).
38杨友生,王耀明,郭世杰.基于小波变换和支持向量机的人脸检测.微机发展.2004年,第14卷(第11期).
39李士进,朱跃龙,王志坚.基于多分类器组合的多角度彩色人脸图像检测.小型微型计算机系统,2004年,第25卷(第8期).
40李军,陈光梦.小波域中基于模糊的图像去噪方法.复旦学报(自然科学版),2005年,第44卷(第6期).
41秦树春,张淑玲,邹复民.基于运动检测和小波分析的视频图像人脸检测.科学技术与工程.2005年,第5卷(第16期).
42龙云利,刘安芝,刘希顺.基于小波变换与支持向量机的虹膜识别新算法.微计算机信息.2005年,第21卷(第12-2期).
43张强,郭宝龙.基于Curvelet变换的图像融合算法.吉林大学学报.2007,第37卷(第2期).
44肖小奎,黎绍发.加强边缘保护的Curvelet图像去噪方法[J].通信学报,2004,(2).
45曹德平.基于支持向量机的彩色人脸检测技术.华侨大学硕士学位论文.2005;40-45.
46杨清夙,游志胜,张先玉.基于豪斯多夫距离的快速多人脸检测算法.电子科技大学学报,2004年04期.
47曹刚,游志胜,刘直芳.基于小波隐性马尔可夫模型的人脸检测.信号处理.2004年,第20卷(第1期).
48倪林,Y.Miao.一种更适合图像处理的多尺度变换——Curvelet变换[J].计算机工程与应用,2004,(28).
49薛明东,郭立,张国宣,刘士建.一种新的图像特征提取算法[J].计算机应用,2004,(S1).
50邓承志,汪胜前,钟华,刘祝华,邹道文.基于Contourlet变换的图像去噪算法[J].电视技术,2004,(10).
51张忠波,马驷良,马捷.小波和神经网络在人脸光照校正中的应用.吉林大学学报(理学版)2005年,第43卷(第2期).
52陈健,周利莉,史红刚,苏大伟.一种基于Haar小波变换的彩色图像人脸检测方法.中文核心期刊《微计算机信息》(测控自动化).2005年第21卷(第10-1期).
53王虹,王秀锋.小波包分析在头肩序列的人脸检测中的应用研究.武汉理工大学学报(交通科学与工程版).2005年,第29卷(第6期).