改良膚色基底AdaBoost臉部偵測系統

全文

(1)AdaBoost. I.

(2)

(3)

(4)

(5)

(6)

(7) (Skin-Based Face Detection Algorithm). I.

(8) ABSTRACT Face detection is widely used in control systems and consumer electronics products. It's an important issue how to save computing resources effectively in face detection. Face detection mainly focused on how to improve the detection rate in the previous research ago. That machine learning methods are used to face detection greatly enhances the face detection rate. Some studies have focused on how to reduce the computing time. In recent years, even machine learning methods and contour texture are used to face detection, detection time is often over-consumptive for feature recognition. Researchers have proposed the use of color to filter the non-face region and to use machine learning algorithms to confirm the candidate region. It effectively reduces the computation time. The aim of our research is to use face information to effectively obtain candidate region. Thus, the face detection time can further reduced. In this research, we find a threshold that determines the accuracy of the detector. Using this threshold to find the most suitable accuracy of the face, it can improve the detection performance.. Keywords : Face Detection, Skin Color Detection, Pattern Recognition.. II.

(9) ................................................................................................................................ I ABSTRACT ................................................................................................................ II ............................................................................................................................. III .......................................................................................................................... V ......................................................................................................................... VI ............................................................................................................... 1 1.1. ................................................................................................ 1. 1.2. ................................................................................................ 4. 1.3. ............................................................................ 6. 1.4. ................................................................................................ 6 ....................................................................................................... 9. 2.1. ................................................................................................ 9. 2.2. .............................................................................................. 11. 2.3 ADABOOST. ............................................................... 14. ..................................................................................................... 17 3.1 3.2 OPENCV. .............................................................................................. 17 .................................................................................. 18. III.

(10) 3.3. .......................................................................................... 18. 3.4. .......................................................................................... 19 ADABOOST. ........................................................ 21. 4.1. .............................................................................................. 21. 4.2. .............................................................................. 23. 4.3 ADABOOST. ............................................... 29. 4.4. .............................................................................. 36. 4.5. .............................................................. 37 ..................................................................................................... 49. 5.1. ...................................................................................... 49. 5.2. .......................................................................................... 51. 5.3. SB-FBA. .................................................................. 51 ................................................................................. 59. .............................................................................................. 60 .............................................................................................. 61. IV.

(11) 1- 1.. 5. 1- 2.. ................................................1. ……... .........................................................................3. 4- 1.. Viola Haar-like. 5- 1.. SB-FDA. 5- 2.. SB-FDA. ……………………………………………..31 PXA270. (sec). ..................................55. (sec). ..............................................................57. V.

(12) 3- 1.. XScale-PXA270 ............................................................................ 17. 4- 1.. .................................................................................... 21. 4- 2.. ................................................................................ 24. 4- 3.. 3x3. ................................................................. 25. 4- 4.. 5x5. ................................................................. 25. 4- 5.. (. 4- 6. 4- 7. 4- 8. 4- 9.. ........................................................................ 29 Viola Haar-like. .............................................................. 30. D. ........................................................... 31. Gentle AdaBoost. 4- 10. 4- 11.. ) ................................................................. 28. ....................................................... 32 .......................................................................... 33. AdaBoost. ................................................................. 34. 4- 12.. .................................................. 37. 4- 13.. SB-FDA. (. )........................... 38. 4- 14.. .......................................... 39. 4- 15.. ...................................................... 40. 4- 16.. .............................................................................. 41. 4- 17.. scale. ................................................................. 42. 4- 18.. scale. ............................................................. 43. VI.

(13) 4- 19.. scale. ..................................................................... 43. 4- 20.. .................................................. 44. 4- 21.. .............................................. 45. 4- 22.. .................................................. 45. 4- 23.. .......................................... 46. 4- 24.. ...................................... 47. 4- 25.. .............................................. 47. 4- 26.. SB-FDA. .............. …………………………...48. 5- 1.. ............................ 49. 5- 2.. ........................................................................ 50. 5- 3.. ........................................................................................ 51. 5- 4.. PC. SB-FDA. AdaBoost. ....................... 52. 5- 5.. PC. SB-FDA. AdaBoost. ........................... 53. 5- 6.. SB-FDA. ........................................... 54. 5- 7.. SB-FDA. ............................................... 54. 5- 8.. PXA270. SB-FDA. SDP-FDA. .................. 56. 5- 9.. PXA270. SB-FDA. SDP-FDA. ...................... 56. 5- 10.. PXA270. SB-FDA. VII. SDP-FDA. ................ 57.

(14) 1.1. (. 2009). 5. 1-1. 2638 36% 64%. (. 945 6953. ). 1- 1.. 5. ( ) 93(2004). 2,836. 964. 33.99%. 94(2005). 3,118. 1,041. 33.39%. 95(2006). 2,880. 1,130. 39.24%. 96(2007). 2,165. 838. 38.71%. 97(2008). 2,193. 753. 34.34%. 2,638. 945. 35.82%. 1.

(15) 123. (. 1981) 1-2 (. 2004). (. (. 74.4%. 2009). 2009). 2. ，.

(16) 1- 2.. 477. 79.4%. 124. 20.6%. 154. 25.6%. 447. 74.4%. 63. 10.5%. 538. 89.5%. 122. 20.3%. 479. 79.7%. 3.

(17) 1.2. (. 2009). (Sun, Kim , & Lee, 2002). Yilmaz Object Tracking: A Survey (occlusions) (Yilmaz, Javed, & Shah, 2006). 4.

(18) (. :. ) (Yang, 2002). (Viola & Jones, 2004). AdaBoost. 5.

(19) 1.3. AdaBoost 1.. AdaBoost. 2.. 1.. 2.. 1.4. AdaBoost. 6.

(20) :. ; (color space). (color detection). (surveillance). AdaBoost. ;. 7. ;.

(21) 8.

(22) 2.1. (. RGB YCbCR. ). CMY HSV(Vezhnevets, Sazonov, & Andreeva, 2004). RGB (. ). Vezhnevets. (Vezhnevets, Sazonov,. & Andreeva, 2004). Nevatia. RGB. (Nevatia, 1977) YCbCr. JPEG MPEG DVD (Y). (Cb Cr) YCbCr. RGB CMY. (. (Vezhnevets, Sazonov, & Andreeva, 2004) ). HSV H. 9. S. V.

(23) HSV. cam-shift (Chu, Ye, Guo, & Liu, 2007). Jo (Pham-Ngoc, & Jo, 2006) Gallardo. (Montufar-Chaveznaza, Gallardo, & Hernandez, 2005) Hsu. (Hsu, Abdel-Mottaleb, & Jain, 2002). 10.

(24) 2.2. (Nefian & Hayes, 1999). Yang (Yang, 2002). 4. 2.2.1. Yang. 3. Huang, 1994). Level. Knowledge-based. Level 1. Level 2 Level 3. rule(Ex.. ). 2.2.2. 11. (Yang &.

(25) Guan (Guan, 2007). (Sun, Kim, & Lee, 2002). (Chai & Bouzerdoum, 2000 ; Phung, Bouzerdoum, & Chai, 2003 ). (Kong & Zhu, 2006). (Tang & Feng, 2008) LBP(Ojala,. Pietikäinen, & Mäenpää, 2002) HOG(Jia & Zhang, 2007). 2.2.3. (Yang, 2002). (Lanitis, Taylor & Cootes, 1995). 2.2.4. 12. SIFT(Geng & Jiang, 2008).

(26) (Viola & Jones, 2004). SVM(Shavers, Li, & Lebby, 2006) AdaBoost(Viola & Jones, 2004 ; Zhu & Zhu, 2008) (Mohamed, Weng, Jiang, & Ipson, 2008) 9 HMM. 5% (. (Nefian & Hayes, 1999) 2.2.5. (Viola & Jones, 2004). 13. HMM ).

(27) 2.3 AdaBoost. (Hidden Markov Model) (Lee, Hon, & Reddy, 1990). (Support Vector Machines) (. 2004). (Neural Networks) (Mohamed, Weng, Jiang, & Ipson, 2008) Lebby, 2006). (Nefian & Hayes, 1999). (Shavers, Li, &. AdaBoost(Viola & Jones, 2004 ;. Zhu & Zhu, 2008) 2004 AdaBoost. Viola. (Viola & Jones, 2004). Boosting 1990. Boosting. Schapire. 14. (Schapire, 1990).

(28) 1995. Freund. Boosting. AdaBoost(Adaptive Boosting). (Freund, 1995) AdaBoost. (adaptively) AdaBoost Viola AdaBoost. (Viola & Jones, 2004). 15. Boosting.

(29) 16.

(30)

(31) Intel Pamium M Process 740. 512MB. OS. Linux kernel 2.6.25. Windows. (OpenCV. Fedora 9. 3.2 OpenCV. OpenCV. Intel C. C++ Linux. , 2006). Matlab. OpenCV. 3.3. CMU PIE. (Gross R, 2000). FERET. (The Color FERET. Database, 2008). FERET. FERET 1993. 1997. FERET. 18.

(32) (The Color FERET Database, 2008) 14051. 8 bit 1208. 3.4. ( Lienhart, Kuranov & Pisarevsky, 2003). AdaBoost. Rainer Lienhart & Pisarevsky, 2003). ( Lienhart, Kuranov David Bradley. AdaBoost. 19.

(33) 20.

(34)

(35) ( 2009). (. ). 。. 22.

(36) 4.2. 。 4.2.1. (Yang, 。. 2002). Yang (intensity). (chrominance) (Yang, 2002) 。. (Hsu, Abdel-Mottaleb, & Jain, 2002). (Ben Hmida & Ben Jemaa, 2006) 。. Ben Hmida. 23.

(37)

(38)

(39) (1). (Erosion) ，. , ，. (Jean Serra. & Pierre Soille, 1994)。. XΘB =. IX. (4-2). y. y∈Bˆ. X. XΘB， X y = {x + y : x ∈ X }. B. Bˆ = {b : −b ∈ B}。. B. Bˆ. X 。. (2). (dilation) , (Jean Serra. & Pierre Soille, 1994)。. X ⊕B=. UX. (4-3). y. y∈B. X. X⊕B， X y = {x + y : x ∈ X } Bˆ. B. B = {b : b ∈ B}。. B ,. A. 。 (3). (opening). 26. X. B. X.

(40) (Jean Serra & Pierre Soille, 1994)。. A o B = ( AΘB ) ⊕ B A. B. A B. (4-4) A. A. B. 。. B. (closing). (4). (Jean Serra & Pierre Soille, 1994)。. A • B = ( A ⊕ B )ΘB A. B. A B. A. 。. B. (4-5). 。. 。. 27. A. B.

(41)

(42)

(43)

(44)

(45)

(46)

(47)

(48) AdaBoost (Freund & Schapire, 1996) 2.. AdaBoost. (. 2009). 3.. (search window). 4.3.5. AdaBoost. 20X20. 4.. (scale). 24X24. 4.3.5. scale. scale 1.1. 1.2. 35.

(49) 4.4. 4-6. AdaBoost AdaBoost. (. 2009) AdaBoost.  FC i = 1, if w > Wsub− window ∩ h > H sub − window  Otherwise  FC i = 0, FCi. w. Wsub-window. (4-7). h. Hsub-window. T. RoSP =. ∑S. i. i. T.  S i = 1, Pixeli ∈ Skin  S i = 0, Otherwise. (4-8). Ratio of Skin Pixel (RoSP) RoSP. T Skin. 36. Pixeli.

(50)

(51)

(52) 4.5.1. (search window). 20x20 24x24. 960X720. 100 95. 90. 85. 80 70. 68. 60 53. 50. (%). 41. 40. 偵測率(%). 30 25. 20. 11. 10 0 20. 6. 0 0 0 72 144 216 288 360 432 504 576 648 720. (pixel) ( & 4- 14. 4-14 576. 5 20X20. 39. 216X216. ).

(53) 1.8 1.6. 1.6. 1.4 1.2 1 0.8. 偵測時間. (s). 0.6 0.4. 0.36 0.23 0.22 0.17 0.14 0.13 0.13 0.12 0.11 0.11. 0.2 0 20. 72 144 216 288 360 432 504 576 648 720. (pixel) ( &. ). 4- 15. 4-15. 72X72. 20X20. 22.5%. 10% (Total Performance Evaluation) (Detection Rate). (Execution time) (TPE) =. (1-. (DR))(. 40. (ET)). (4-10).

(54) 14 12 10.3410.0310.5. 10 8. 8. 6. 12.22 11.57 11. 11. 11. 7.36 5.4. 4 2 0 20. 72 144 216 288 360 432 504 576 648 720. (pixel)( &. ). 4- 16. 4-15. 72x72. 72x72 4.5.2. (scale). scale (. scale 2008). scale. scale. 4-17. 41. 1.25.

(55) 100 90 80 70 60. 95 87 78 73 67. 64 64. (%). 51. 50. 45 45. 40 30. 48. 28. 20. 31 24. 20. 17 12 11. 10 0. 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2 2.5 3 3.5 4 4.5 5 10 20. scale 4- 17. scale 4-17. scale scale. scale. 42.

(56) 1.6 1.49 1.4 1.2 1 0.91. (s). 0.8 0.6. 0.6 0.5. 0.4 0.2. 0.450.45 0.370.370.370.38 0.32 0.230.270.260.230.240.270.23. 0 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2 2.5 3 3.5 4 4.5 5 10 20. scale 4- 18.. scale. 4-18. scale. 30 25. 24.03 21.12 20.9 20.52 20.35 19.09 18.4 18.13 17.94 16.64 16.56. 20 15. 16.2 14.85 13.5 13.32 13.2 11.83. 10 7.45 5 0 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2 2.5 3 3.5 4 4.5 5 10 20. scale 4- 19.. scale. 43.

(57) 4-19. scale. scale. 1.1. scale. 4.5.3. scale. scale. 1.1. (960X720). 1/10. 98 96 94.6. 94. 94.6. 95.4. 95.4. 95.4. 95.4. 95.4. 95.4. 95.4 95.4. 92 90. (%). 88 86 85. 84 82 80 78 0. 69120 138240 207360 276480 345600 414720 483840 552960 622080 691200. (pixel) 4- 20. 4-20. 72X72 1/10 72X72. 44.

(58) 1.6 1.42. 1.4 1.2. 1.2. 1.26. 1.09 1 0.87. 0.8. 0.93. (s). 0.69 0.6. 0.59. 0.62. 0.45. 0.4 0.32 0.2 0 0. 69120 138240 207360 276480 345600 414720 483840 552960 622080 691200. (pixel) 4- 21.. 7 6.532 6 5. 5.4. 5.14. 4. 4.002 3.186. 3. 2.852. 3.174. 2.43 2 1 0. (pixel) 4- 22.. 45. 4.278. 5.52. 5.796.

(59) 4-22. 1/10. 1/10 scale 4.5.4 (Yang, 2002). 96 94.6. 94. 94.6 92.4. 92. 90.1. 90. 90.1. 90.1. 90.1. (%). 89.3 88.5. 88. 87.8 86.25. 86 84 82 0. 10. 12. 14. 16. 18. 20. 25. ( ) 4- 23.. 46. 30. 35. 40.

(60) 50 45. 45. 43. 41. 40. 38. 37. 36. 35. 36. 35 31. 30. 30. 28. 25. (s). 20 15 10 5 0 0. 10. 12. 14. 16. 18. 20. 25. 30. 35. 40. ( ) 4- 24. 4-23. 4.5 4 3.76 3.5. 3.56. 3.663 3.56. 3.75. 3.57. 3.66. 3.85. 3.12. 3 2.5. 2.43. 2.32. 2 1.5 1 0.5 0 0. 10. 12. 14. 16. 18. 20. ( ) 4- 25.. 47. 25. 30. 35. 40.

(61)

(62) SB-FDA XScale-PXA270. 5.1. XScale-PXA270. PC. 300 250 200 一般. 150. (ms). 查表法. 100 50 0 640*480. 960*720. 1280*1024. 5- 1. 5-1. PC. 49.

(63)

(64)

(65) M6B00V. PXA270 SB-FBA. SB-FBA. 5.3.1. M6B00V. AdaBoost. SB-FBA 3 2.5. 2.45. 2 1.65. 1.5. 傳統AdaBoost. (s). 膚色基底AdaBoost 1.02. 1. 0.66 0.5 0 640x480. 5- 4.. PC. 960x720. SB-FDA. AdaBoost. 5-4 SB-FDA. 52.

(66) 95.4. 96 94. 93.1. 92 傳統AdaBoost. 90. (%). 88.5. 膚色基底AdaBoost臉部偵測. 87.8. 88 86 84 640x480. 5- 5.. 960x720. PC. SB-FDA. AdaBoost. AdaBoost SB-FDA. SB-FDA AdaBoost. 53. AdaBoost.

(67) 3 2.5. 2.45. 2 1.65. 1.5. 傳統AdaBoost SB-FDA. (s). 1.02. 1. Improve SB-FDA. 0.66 0.5. 0.45 0.18. 0 640x480. 960x720. 5- 6.. SB-FDA. 5-6. improve SB-FDA. 95.4. 96. 94.6 94. 93.1 93.1. 92 傳統AdaBoost. 90. (%). 88.5. SB-FDA 87.8. 88. Improve SB-FDA. 86 84 640x480. 5- 7.. 960x720. SB-FDA. 54.

(68) improve SB-FDA. SB-FBA. 5.3.2. 5.3.1. SB-FDA. PXA270. M6B00V. SB-FBA SB-FBA. 5- 1 T-AFDA. PXA-270. AdaBoost. AdaBoost 5- 1.. SB-FDA. PXA270. (sec).. 640x480. 960x720. T-AFDA. 40.42. 101.01. SB-FDA. 34.5012. 99.1563. 9.454. 21.26. SB-FDA. 55. SDP-FDA.

(69) 120 101 100. 99. 80 T-AFDA. 60. SB-FDA. (s). 40. 改良式 SB-FDA. 34. 40. 21 20. 9. 0 640x480. 5- 8.. 960x720. PXA270. SB-FDA. SDP-FDA. 95.4. 96. 94.6 94. 93.1 93.1. 92 傳統AdaBoost. 90. (%). 88.5. SB-FDA 87.8. 88. Improve SB-FDA. 86 84 640x480. 5- 9.. 960x720. PXA270. 5- 8. SB-FDA. SDP-FDA. PXA270 SB-FDA. SB-FDA SB-FDA. 56.

(70) 5- 2.. SB-FDA. (sec).. 640x480. 960x720. T-AFDA. 1. 1. SB-FDA. 1.174. 1.018. 4.2754. 4.751. SB-FDA. 4.751. 5 4.5. 4.2754. 4 3.5 3 SB-FDA. 2.5. 改良式 SB-FDA. 2 1.5. 1.174. 1.018. 1 0.5 0 640x480. 5- 10.. 960x720. PXA270. SB-FDA. 5- 10 SB-FDA. SDP-FDA SB-FDA. 4.5. 57.

(71) 58.

(72) AdaBoost. AdaBoost. 59.

(73) (2009) ：http://www.tcpd.gov.tw/upload/attachment/ e209980a4373b498e0 f15bbd82bcb.pdf. (1981). (2004). ,. (2009). (2004) OpenCV. SVM. 1-7. (2006) OpenCV ：http://www.opencv.org.cn/index.php/OpenCV%E6%A6%82%E8%BF%B0. (2008). Face Detection Based on AdaBoost. 60.

(74) Vezhnevets, V., Sazonov, V., & Andreeva, A. (2004). A Survey on Pixel-Based Skin Color Detection Techniques, GraphiCon-2003, pp.85-92.. Nevatia, R. (1977). A color edge detector and its use in scene segmentation, IEEE Transactions on Systems, Man and Cybernetics, Vol: 7, pp.820-826.. Chu, H., Ye, S., Guo, Q., & Liu, X. (2007). Object Tracking Algorithm Based on Camshift Algorithm Combinating with Difference in Frame, IEEE Automation and Logistics, pp.51-55.. Pham-Ngoc, P. T., & Jo, K. H. (2006). Color-based Face Detection using Combination of Modified Local Binary Patterns and embedded Hidden Markov Models, SICE-ICASE, pp.5598-5603.. Montufar-Chaveznaza, R., Gallardo, F. H., & Hernandez, S. P. (2005). Face detection by polling, Intelligent Signal Processing, 2005 IEEE International, pp.292-297.. Hsu, R. L., Abdel-Mottaleb, M., & Jain, A. K. (2002). Face Detection in Color Images,Pattern Analysis and Machine Intelligence, IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol:24, pp.696-706.. Yang, M. H. (2002). Detection Faces In Image : A Survey,Pattern analysis and Machine, IEEE Transactions on Pattern Analysis and Machine Intelligence,. Vol:24, pp.34-58.. Yang, G., & Huang, T. S. (1994). Human Face Detection in Complex Background, Pattern Recognition, vol. 27, pp. 53-63.. 61.

(75) Guan, Y. (2007). Robust Eye Detection from Facial Image based on Multi-cue Facial Information, IEEE Control and Automation, pp.1775-1778.. Sun, Y. B., Kim, J. T., & Lee, W. H. (2002). Extraction of face objects using skin color information, IEEE Communications, Circuits and Systems and West Sino Expositions, vol.2, pp.1136-1140.. Chai, D., & Bouzerdoum, A. (2000). A Bayesian approach to skin color classification in YCbCr color space, TENCON, vol.2, pp.421-424.. Phung, S. L., Bouzerdoum, A., & Chai, D. (2003). Skin segmentation using color and edge information, ISSPA, vol.1, pp.525-528.. Kong W., & Zhu, Shan'an. (2006). A New Method of Single Color Face Detection Based on Skin Model and Gaussian Distribution, WCICA, vol.1, pp.261-265.. Tang, H. K., & Feng, Z. Q. (2008). Hand's Skin Detection Based on Ellipse Clustering, ISCSCT, vol.2, pp.758-761.. Ojala, T., Pietikäinen, M., & Mäenpää, T. (2002). Multiresolution gray-scale and rotation invariant texture classification with local binary patterns, IEEE Pattern Analysis and Machine Intelligence, vol.24, pp.971-987.. Jia, H. X., & Zhang, Y. J. (2007). Fast Human Detection by Boosting Histograms of Oriented Gradients, IEEE Image and Graphics, pp.683-688.. Geng, C., & Jiang, X. (2009). SIFT features for face recognition, ICCSIT, pp.598-602.. Han, C. H., & Sim, K. B. (2008). Real-time face detection using AdaBoot algorithm, ICCAS, pp.1892-1895.. 62.

(76) Shavers, C., Li, R., & Lebby, G. (2006). An SVM-based approach to face detection, SSST, pp.362-366.. Viola. P., & Jones, M. J. (2004). Robust real-time face detection, International Journal of Computer Vision, vol.57, pp.137-154.. Zhu, H., & Zhu, S. (2008). Face detection based on AdaBoost algorithm with differential images, ICALIP, pp.718-722.. Mohamed, A., Weng, Y., Jiang, J., & Ipson S. (2008). Face detection based neural networks using robust skin color segmentation, IEEE SSD, pp.1-5.. Nefian, A. V., & Hayes III M. H. (1999). Face recognition using an embedded HMM, IEEE Conference on Audio and Video-based Biometric Person Authentication, pp.15-19.. Yilmaz, A., Javed, O., & Shah, M. (2006). Object Tracking: A Survey, ACM Computing Surveys (CSUR), vol:38, pp.1-45.. Lee, K. F., Hon H. W., & Reddy, R. (1990). An overview of the SPHINX speech recognition system, IEEE Acoustics, Speech and Signal Processing, vol:38, pp.35-45.. Schapire, R. E. (1990). The Strength of Weak Learnability, Foundations of Computer Science, vol: 5(2), pp.197-227.. Lienhart, R., & Maydt, J. (2002). An Extended Set of Haarlike Features for Rapid Object Detection, IEEE International Conference on Image Processing, Vol.1, pp.900-903.. Freund, Y. (1995). Boosting a Weak Learning Algorithm by Majority, Information and Computation, vol:2, pp.256-285.. Ben Hmida, M., & Ben Jemaa, Y. (2006). Fuzzy classification, image segmentation and shape analysis for Human face detection, ICECS '06. 13th, pp.640-643.. 63.

(77) Yang, M. H., & Ahuja, N. (1999). Gaussian Mixture Model for Human Skin Color and Its Applications in Image and Video Databases, SPIE ’99, pp.458-466.. Kovac, J., Peer. P., & Solina, F. (2003). Human Skin Color Clustering for Face Detection, Proceedings of the IEEE Region 8 Computer as a Tool, Vol.2, pp. 144-148.. Lanitis, A. C., Taylor, J., & Cootes, T. F. (1995). An Automatic Face Identification System Using Flexible Appearance Models, Image and Vision Computing, vol. 13, no. 5, pp. 393-401.. Serra, J., & Soille, P. (1994). Mathematical morphology and its applications to image processing, Kluwer Academic Publishers, Boston.. Chang, F. C., Chen, J., & Lu, C. J. (2004). A Linear-Time Component-Labeling Algorithm Using Contour Tracing Technique, Computer Vision and Image Understanding, Vol. 93, pp. 206-220.. Freund, Y., & Schapire, R. E. (1996). Experiments with a New Boosting Algorithm, Proceedings of the Thirteenth International Conference on In Machine Learning, Morgan Kauman, San Francisco, pp. 148-156.. Gross R. (2000). PIE Database [Online forum comment]. Retrieved from http://www.ri.cmu.edu/research_project_detail.html?projec t_id=418andmenu_id=261. The Color FERET Database (2008). The Color FERET Database[Online forum comment]. Retrieved from http://face.nist.gov/colorferet/. Lienhart, R., Kuranov, A., & Pisarevsky, V. (2003). Empirical Analysis of Detection Cascades of Boosted Classifiers for Rapid Object Detection, DAGM-Symposium, pp. 297–304.. 64.

(78)