21 Edge Feature for Monochrome Image Retrieval 1100311 2010 3 1
3 3 2 2 7 200 Sobel Canny i
Abstract Edge Feature for Monochrome Image Retrieval Naoto Suzue Content based image retrieval (CBIR) has been proposed as one of the image retreival system. The user imagines shape of the image and inquire the image retrieval system. The sketch image retrieval using edge feature has been proposed. The sketch image retrieval has a problem that the result of the image retrival is different depending on the user. Moreover, because the edge is the local feature, edge of the entire image cannot used for understanding position of edge. Conventional CBIR system are mostly designed for color image retrival. Then, the user imagines a shape feature and a color feature. Therefore, verification of retrieval performance of the image retrieval for edge features is difficult. Then we propose the combination of local features and golbal features. The local feature is dege of segmented region of image. The global feature is difference of average brigthness of areas of a image. 200 images and seven testees are used in the experiment. And, Sobel filter, Canny filter and proposal technique are used in the experiment. The results of the image retrieval indicate that accuracy of retrieval of the proposal technique improved more than the experiment key words monochrome image retrieval, edge feature, division image, local feature, global feature ii
1 1 2 4 2.1.................................. 4 2.2 1............................... 4 2.2.1 Sobel................................ 6 2.2.2 Canny............................... 8 3 12 3.1.............................. 12 3.2.............................. 13 3.3............................. 14 4 16 4.1.......................... 16 4.2................................... 16 4.3................................. 20 4.4................................... 21 4.5............................. 24 5 26 28 30 A 32 iii
2.1................. 5 2.2................................ 5 2.3 Sobel................................. 6 2.4 8 Sobel d........................ 7 2.5..................................... 9 2.6....................... 10 2.7......................... 10 2.8 Sobel......................... 11 2.9 Canny......................... 11 3.1............................ 15 3.2............................ 15 4.1............................ 18 4.2................................. 19 4.3................................. 20 4.4..................................... 22 4.5..................................... 22 4.6..................................... 23 4.7..................................... 23 A.1........................ 32 A.2........................ 33 iv
4.1................................ 17 4.2................ 21 4.3............................... 21 v
1 2 TBIR (Text Based Image Retrieval) [1][2][3] [4][5][6][7][10][11][12] CBIR (Content Based Image Retrieval) TBIR TBIR CBIR CBIR 2 1
[7][10] [10]. [4][12] [4] [12], 2 1 2
3 3 2 2 2 2 2 3 4 5 3
2 2.1 1 ( ) 2 ( ) 1 Sobel Canny 2.2 1 x = f(x + 1, y) f(x, y) (2.1) y = f(x, y + 1) f(x.y) (2.2) x y 2.6 2.6 4
2.2 1 2.1 2.1 2.2 2.7 2.7 2.1 2.2 5
2.2 1 2.2.1 Sobel Sobel Sobel 2.3 2.3 Sobel 2.4 8 Sobel [7] Sobel HSV H S V Sobel S V f Sd (x, y) f Id (x, y) d(d 0, π/4, π/2, 3π/4, π, 5π/4, 3π/2, 7π/4, 2π) e d W H e d = 1 W H W 1 x=0 H 1 y=0 G(f Sd (x, y), r S ) G(f Id (x, y), r I ) (2.3) G(a, b) = { 1 if (a < b) 0 otherwise (2.4) r S = τ S max f S d(x, y) (2.5) r I = τ I max f S d(x, y) (2.6) τ S 0.35 τ I 0.15 8 2.8 6
2.2 1 2.4 8 Sobel d 7
2.2 1 2.2.2 Canny Canny [9] Sobel Canny. Sobel 0 2 68 128 2.9 8
2.2 1 2.5 9
2.2 1 2.6 2.7 10
2.2 1 2.8 Sobel 2.9 Canny 11
3 2 3.1 Canny 3 3 300 300 3.1 3.1 12
3.2 1. 2. Canny 2.8 2.9 Sobel Canny 2 3. 3 3 4.. 0 1 9 = (3.1) 3.2 2 2 13
3.3 2 2 3.2 3.2 1. 2 2 2.. 3.2 = (3.2) 3.3 0 1 255 255 0 1 14
3.3 3.1 3.2 15
4 Sobel Canny 4.1 4.2 3 1 Sobel 2 Canny 3 3 2 4.1 ArtExplosion 10 20 16
4.2 4.1 15 Sobel 8 Canny 1 200 ArtExplosion 10 1 4.2 10 4.2 200 7 4.1 N C N C R = R N (4.1) = R C (4.2) 17
4.2 4.1 18
4.2 4.2 19
4.3 4.3 1 4.3 4.3 1. 1 2. 200 3. 200 4. 1 20 4.3 20
4.4 4.4 4.2 4.2 4.4 4.5 4.6 4.7 4.3 4.2 1 2 3 4 5 6 7 8 9 10 1 8 9 9 16 10 11 5 7 8 4 2 15 8 5 7 4 16 4 9 9 6 3 7 8 9 9 3 16 6 8 11 3 4 2 2 2 3 2 3 3 3 2 3 5 9 4 1 2 2 6 4 6 4 3 6 3 6 3 4 2 7 2 2 1 2 7 1 7 4 3 3 7 2 2 1 2 4.3 13 57 Sobel 8 40 Canny 10 44 21
4.4 4.4 4.5 22
4.4 4.6 4.7 23
4.5 4.5 3 Sobel Canny Sobel Canny Sobel Canny Sobel Canny 4.2 10 2 20 2 7 9 24
4.5 25
5 Canny 3 3 2 2 13 57 3 14 2 2 3 3 4 4 26
27
PC Open CV 1 28
FreeBSD PC to!ht. 3 1 2 29
[1],,,,, vol. 54, No. 11, pp. 1631 1638, 2000. [2],,, HTML,, vol. J87-D-I, no. 2, pp. 145 153, 2004. [3],,,,, vol. J82-D-II, no. 11. pp. 2178 2179, 1999. [4],, Canny,, vol. 127, no. 11, pp. 1880 1887, 2007. [5], 20 [6],, 12 [7] M. Serata, Y. Hatakeyam, and K. Hirota, Designing Image Retrieval System with the Concept of Visual Keys, J. of Advanced Computational Intelligence and Intelligent Informatics, vol. 10, no. 2, pp. 136 144, 2006 [8],, pp. 105 121, 2004/7/22 [9] J. Canny, A Computational Approach To Edge Detection. IEEE Trans. Pattern Anylysis and Machine Intelligence, vol. 8, no 6, pp. 679 698, 1986. [10],, 20 [11],,, SOM, ] 13 (DEWS 2002)B1-3, 2002 30
[12],,,,,, 56, 4, pp. 653 658, 2002. 31
A Canny 2 2 3 3 4 4 4 A.1 32
A.2 33