28 TCG SURF Card recognition using SURF in TCG play video 1170374 2017 3 2
TCG SURF TCG TCG OCG SURF Bof 20 20 30 10 1 SURF Bag of features i
Abstract Card recognition using SURF in TCG play video Haruka MOTOZONO Today, TCG s play video has been distributed with the spread of the Internet and video distribution services, and many people have enjoyed the game situation. However, there are a lot of kinds of TCG cards. In addition, texts on tha cad for card names and explanation are small. Therefore it is difficult to understand which cards are used in TCG play video. In this research, we propose card recognition using SURF in video presenting cards for Yu-Gi-Oh! OCG The processing is as follows, 1)extract the card from each frame of the video using contour extraction, 2)identify the card type, 3)create a histogram by Bof, 4)compare the histogram of the registered templates for the specfied card type, 5)use the top 20 template images with high similarity as candidates, 6)compare the distance between the corresponding feature points, 7)choose smallest distance one as a result of card recognition. In the experiments, we selected 10 cards from each card types randomly, and examined whether it is recognized correctly by using this proposed method. In addtion, we evaluated average processing time per frame. key words Trading Card Game SURF Bag of features ii
1 1 1.1................................... 1 1.2................................. 2 2 3 2.1 (TCG).................... 3 2.2 SURF................................. 4 2.3 Bag-of-features............................... 5 3 7 3.1................................... 7 3.2................................ 9 3.3............................. 10 3.4............................. 12 3.5 SURF........... 13 3.6................................. 14 3.7............................. 15 3.8................................. 15 4 16 4.1................................ 16 4.2................................... 16 4.3................................... 17 4.3.1..................... 17 4.3.2............................. 21 iii
4.4...................................... 22 5 24 25 26 iv
2.1 Bag-of-features............................ 5 2.2 visualwords............................. 6 2.3............................. 6 3.1............................... 8 3.2............................. 9 3.3 ( ) ( ) ( )..... 10 3.4...... 11 3.5... 11 3.6 ( ) ( ) 180 ( ).................................... 13 3.7............................... 14 3.8 ( ) ( )........ 15 4.1 1........ 17 4.2......................... 17 4.3............................... 18 4.4 1..................... 23 v
3.1............ 8 3.2............................. 12 4.1............................... 18 4.2...................... 19 4.3.................... 20 4.4......................... 20 4.5........................ 21 4.6................. 21 4.7 [sec]......... 21 vi
1 1.1 2 [1] TCG WIXOSS( ) TCG TCG ( OCG)[2] SURF [3] 1
1.2 1.2 SURF Bag of features TCG SURF 2
2 SURF Bag of features 2.1 (TCG) ( TCG) (CCG) (CCG) 2 TCG TCG TCG 3
2.2 SURF 2.2 SURF SURF(Speeded Up Robust features) SIFT ( ) Hessian-Laplace Box Hessian-Laplace (2.1) [ Lxx (x, σ) L xy (x, σ) H(x, σ) = L xy (x, σ) L yy (x, σ) ] (2.1) (2.1) x, y, σ 3 SURF (2.2) Box (2.3) det(h) = L xx (X, σ) L yy (x, σ) (L xy (x, σ)) 2 (2.2) det(h) = D xx D yy (D xy 0.9) 2 (2.3) Haar-like 20 20 4 4 Haar-like( 2 2) x, y d x, d y 4 v = ( d x, d y, d x, d y ) 16 4 64 [3][4] 4
2.3 Bag-of-features 2.3 Bag-of-features Bag-of-features( Bof) 1 Bof 2.1 DB visual words DB visual words DB [5] 2.1 Bag-of-features visual words 2.2 DB 2.2 kmeans 2.3 DB visual words 5
2.3 Bag-of-features 2.2 visualwords 2.3 6
3, TCG SURF,. 3.1 370 (708 1038pixel) OCG 1/4 1920 1080pixel 30fps OCG 7 6 600mm 510mm 82cm OCG 3.1 9 10 3.1 3.1 7
3.1 3.1 3.1 1 14 17 10 134 2 29 19 98 46 8
3.2 3.2 3.2 3.2 SURF SURF visual words Bof Bof 20 3.2 9
3.3 3.3 3.3( ) 3.3( ) 3.3( ) HSV 1/6 1/25 1/3 2/3 3 3pixel HSV 20 HSV 3.2 7 3.2 3.2 HSV 3.4 3.5 3.3 ( ) ( ) ( ) 10
3.3 3.4 3.5 11
3.4 3.2 40 65 16 16 25 105 120 120 145 25 80 100 ( ) 80 105 145 170 3.4 SURF kmeans visual words 256 visual words visual words 12
3.5 SURF 3.5 SURF 3.8 SURF 3.6( ) 3.6( ) 1/7 2/7 180 3.6 180 3.6 ( ) ( ) 180 ( ) 13
3.6 3.6 3.7( ) 3.7( ) 3.7( ) 4 3.7( ) 3.7( ) SURF 3.7 14
3.7 3.7 Ha Hb X 3.1 20 i Min(Ha[i], Hb[i]) X = i Ha[i] (3.1) 3.8 20 SURF SURF 20 3.5 180 3.5 3.8 3.8( ) 180 30 3.8 ( ) ( ) 15
4,,.. 4.1 python. OpenCV scikit-learn numpy 4.2 370 10 10 10 1 2 1 2 4.1 A B C D 1 16
4.3 4.1 1 4.3 4.3.1 10 4.1 4.2 4.3 4.2 17
4.3 4.1 32 32 0 33 278 253 25 254 279 279 0 288 320 309 11 319 290 278 12 284 51 51 0 56 298 273 25 287 298 273 25 284 284 267 17 273 277 260 17 266 4.3 18
4.3 10 4.2 6 4.7 4.4 4.5 4.2 (%) 32 32 0 100 253 253 0 100 279 270 9 96.8 309 242 67 79.3 278 265 43 87.8 51 51 0 100 280 279 1 99.6 306 102 204 43.5 267 220 47 100 260 260 0 96.1 19
4.3 4.3 3 0 6 64 0 0 31 3 0 1 0 0 173 0 0 22 25 0 4.4 (%) 32 32 0 100 253 253 0 100 279 270 9 96.8 309 242 67 79.3 278 265 43 87.8 51 51 0 100 280 279 1 99.6 306 102 204 43.5 267 220 47 100 260 260 0 96.1 20
4.3 4.5 3 26 64 0 31 18 1 0 173 0 22 28 4.7 4.6 4.6 (%) 587 564 23 96.1 587 550 37 98.7 4.3.2 4.1 A B C D 1 4.7 [sec] A B C D [sec] 0.17 0.22 0.28 1.21 1.89 21
4.4 4.4 HSV SURF Bof 370 4.4 (4.1) (4.5) (4.1) A (4.2) B (4.3) C (4.4) D (4.5) X(t) + nh(t) + 20K(t) (4.1) X(t) + nk(t) (4.2) X(t) + nt (t) (4.3) Y (t) + mk(t) (4.4) Y (t) + mt (t) (4.5) 22
4.4 X(t) Y (t) H(t) 1 K(t) 1 T (t) 1 n m 4.4 1 23
5 TCG SURF 1 24
OCG 4 6 3 25
[1] TCG vol. 115 no. 405 pp. 13-18 2016 [2] / /KONAMI http://www.yugioh-card.com/japan/ [3] H.Bay A.Ess T.Tuytelaars and L.Gool Supeededup robust features(surf) Computer Vision and Image Understanding 110 pp. 346-359 2008 [4] 2 pp. 26-28 2010 [5] 3 pp. 63-95 2010 26