20 Method for Recognizing Expression Considering Fuzzy Based on Optical Flow 1115084 2009 3 5
3.,,,.., HCI(Human Computer Interaction),.,,.,,.,.,,..,. i
Abstract Method for Recognizing Expression Considering Fuzzy Based on Optical Flow Masashi Onishi When humans communicate to each other, a face has three roles. They are the roles as the display to express the person s individuality, the indicator to reflect the state of mind at that time and the messenger to express the intention which should be transmitted. The role of face as a media which conveys feelings especially by expression is important. For this reason, it is an important subject for a computer to recognize expression especially in the field of HCI (human computer interaction). In recent researches, it has been researched to divide a subject s expression into some categories and recognize expression. However, since the strength in expression exists, it is necessary to take it into consideration, because the role of more natural media can be played by the ability to recognize the strength of expression. In this research, after describing the extraction method of features for recognizing expression and the classification technique of expression, the technique for showing the strength of expression is proposed. An optical flow is used as features. Functions showing the strength of expressions are calculated. The strength of expressions are shown using the function. key words Strength of Expression, Optical flow, Fitting ii
1 1 1.1................................... 1 1.1.1............................... 1 1.1.2............................ 2 1.1.3 FACS....................... 2 1.2.................................. 3 1.3................................. 4 2 5 2.1............................ 5 2.2.......................... 5 2.3.......................... 7 3 8 3.1................................. 8 3.2........................... 12 3.3........................... 13 3.4................................... 14 3.5.............................. 15 3.6................................... 15 4 16 4.1................................. 16 4.2................................. 28 iii
5 30 6 32 33 35 A 36 iv
2.1.............................. 6 3.1.................................... 8 3.2.................................... 9 3.3................................ 9 3.4................................ 11 3.5................................... 13 3.6.......................... 14 4.1 ()............................. 20 4.2 1 ()................................ 21 4.3 ()............................. 21 4.4 ()............................. 22 4.5 1 ()................................ 22 4.6 ()............................. 23 4.7 ()............................. 23 4.8 1 ()................................ 24 4.9 ()............................. 24 4.10 ()........................... 25 4.11 1 ()............................... 25 4.12 ()........................... 26 4.13........................ 27 v
3.1................................. 9 3.2.......................... 10 3.3...................................... 10 3.4.............................. 10 3.5................................... 12 4.1 ()................................ 16 4.2 ()................................ 17 4.3 ()................................ 17 4.4 ()............................... 18 4.5.................................... 18 4.6............................... 19 4.7............................... 19 4.8............................... 19 4.9.............................. 19 4.10............................ 20 4.11.................................... 28 4.12............................ 28 4.13.......................... 29 A.1 1........................... 36 A.2 2........................... 37 A.3 1........................... 38 A.4 2........................... 39 vi
A.5 1........................... 40 A.6 2........................... 41 A.7 1......................... 42 A.8 2......................... 43 vii
1 1.1 1.1.1., face-to-face comunication,.,, 7%, 38%, 55% Mehrabion, () [1].,,., [2].,, [3]., 1
1.1 [4]. 1.1.2,, [2].,,,,.,. 1.,,,.,,,,. 2.,,.,,. 1.1.3 FACS,,., 2
1.2 Ekman Friesen Facial Action Coding System(FACS)[5],. FACS Action Unit (AU) 46,, AU., AU, AU. AU., AU,, 1.2 1, 2., FACS,,,., FACS FACS. AU [6][7],, AU., FACS AU. 1.2,, [2][8].,, [9].,.., FACS AU 3
1.3,.,,,,. 1.3 2,,,.,,. 3.,,,,.,,. 4, 3.,. 5, 6 4
2 2.1,.,,.,,.,,,. 2.2,.,. 2.1. 5
2.2 2.1 n x i., y i. (x i, y i ).,,, y = 1 ( 2.1), 1 ( 2.2), y = 1 ( 2.3),. 6
2.3 f(x) = 1 1 + exp[ (x 2 )] (2.1) f(x) = ax + b (2.2) f(x) = 1 exp[ x] (2.3),,. 2.3 n x i, n (x i, y i ), n., (2.1),, (2.2) a, b, (2.3)., [10],., (2.4) J. J = n [f(x i ) y i ] 2 (2.4) i=1 7
3 3.1 21.,,,, 4.. 3.1, 3.2. 3.1. 8
3.1 3.2 3.1 Victor GZ-MG77 3.3 9
3.1 3.2 Victor GZ-MG77 1/3.9 218 CCD 123 200 F1.22.0 f=3.838mm(35mm 35.8457mm) MPEG-2 Dolby Digital NTSC HDD, SD 3.3 SLIK ABLE 400 DX-LE 1.55m, 0.33m 2.48kg 3.4 1mm/m=0.0527 5.25mm/m= 0.2999 10
3.1,,, 12.,, 3.4. 3.4 11, 10. 11
3.2 3.2 14., 05 6. 0, 5. 05 3.5. 3.5 5 4 3 2 1 0. 12
3.3 3.3,,. 3.5. 3.5 13
3.4. 10, 30. 3.6. 3.6,. 3.4. n. n, (2.4) J. 14
3.5 3.5... 3.6,,.,. 0. 15
4 4.1 n 4.1 4.4. n 4.5. 4.1 n () 1 0.287 5 0.489 10 0.559 20 0.632 50 0.686 100 0.704 200 0.706 400 0.650 0.632 16
4.1 4.2 n () 1 0.303 5 0.399 10 0.401 20 0.438 50 0.513 100 0.577 200 0.636 400 0.658 0.657 4.3 n () 1 0.634 5 0.713 10 0.727 20 0.751 50 0.770 100 0.771 200 0.758 400 0.729 0.700 17
4.1 4.4 () n 1 0.400 5 0.438 10 0.470 20 0.541 50 0.584 100 0.576 200 0.529 400 0.474 0.458 4.5 n 200 0.706 100 0.771 400 0.658 50 0.584 n,. 4.6 4.9. 18
4.1 4.6 (2.1) = 0.07 = 42.0 (2.2) a = 0.02 b = 0.06 (2.3) = 0.04 4.7 (2.1) = 0.06 = 23.0 (2.2) a = 0.01 b = 0.39 (2.3) = 0.06 4.8 (2.1) = 0.04 = 54.0 (2.2) a = 0.01 b = 0.22 (2.3) = 0.03 4.9 (2.1) = 0.03 = 92.0 (2.2) a = 0.01 b = 0.06 (2.3) = 0.02 19
4.1 4.10, 4.1 4.12. 4.10 (2.1) (2.2) (2.3) 0.134 0.124 0.147 0.155 0.157 0.162 0.194 0.200 0.243 0.111 0.105 0.091 4.1 () 20
4.1 4.2 1 () 4.3 () 21
4.1 4.4 () 4.5 1 () 22
4.1 4.6 () 4.7 () 23
4.1 4.8 1 () 4.9 () 24
4.1 4.10 () 4.11 1 () 25
4.1 4.12 () 26
4.1.,,, 0 0.,, y = 1,, (2.3) ( 4.13). 4.13 27
4.2 4.2,. 10,,, 30. 4.11, 4.12, 4.13. 4.11 66.67% 40.00% 26.67% 86.67% 4.12 80.00% 70.00% 50.00% 70.00% 20.00% 30.00% 40.00% 20.00% 20.00% 70.00% 90.00% 100.00% 28
4.2 4.13 A 66.67% 66.67% 0.00% 66.67% 50.00% B 33.33% 0.00% 33.33% 100.00% 41.67% C 33.33% 66.67% 66.67% 66.67% 58.33% D 66.67% 100.00% 33.33% 100.00% 66.67% E 66.67% 33.33% 0.00% 100.00% 50.00% F 66.67% 33.33% 0.00% 66.67% 41.67% G 100.00% 0.00% 33.33% 66.67% 50.00% H 33.33% 0.00% 100.00% 100.00% 58.33% I 100.00% 33.33% 0.00% 100.00% 58.33% J 100.00% 33.33% 0.00% 100.00% 58.33%, A.1 A.8 29
5,,, ( 4.10).,,, 0 0.,, y = 1,, (2.3) ( 4.13).,., f(x) y i, x i,. f(x) y i, x i,.,., 4.12,,,.,,.,,,., 4.13.,,., 26.67%,, 30
.,,. 31
6,,,,.,,.,,,,.,, 4,,.,.,,,,,,. 32
,,., 3, 4,,,,,.,,,.,.,.,,,,,.,.,,,., 4,., 4.,,,.,,.,, 2,, 1,,,, 4,,,,,,,,,,,,, 3,,,,., 33
,,.,,..,,,.... 34
[1], [], Vol.85 No.9, pp680-685 (2002). [2], [], Vol.85 No.10, pp766-771 (2002). [3], [], Vol.85 No.12, pp936-941 (2002). [4], [], Vol.86 No.1, pp54-61 (2003). [5] P. Ekman and W. V. Friesen, Pictures of Facial Affect, Human Interaction Laboratory, Univ. of California Medical Center, San Francisco, (1976). [6] G. Donato, M. S. Bartlett, J. C. Hager, P. Ekman, and T. J. Sejnowski,Classifying Facial Actions, IEEE Trans. PAMI, Vol.21, No.10, Oct. (1999). [7] Y. Tian, T. Kanade, and J. F. Cohn, Recognizing Action Units for Facial Expression Analysis, IEEE Trans. PAMI, Vol.23, No.2, Feb. (2001). [8],,,, J81-D-2, No.6 pp1150-1159 (1998). [9],,, 20, pp325. [10],,, (2004). 35
A A.1A.8,. A.1 1 A B C D E 36
A.2 2 F G H I J 37
A.3 1 A B C D E 38
A.4 2 F G H I J 39
A.5 1 A B C D E 40
A.6 2 F G H I J 41
A.7 1 A B C D E 42
A.8 2 F G H I J 43