(MIRU2011) 2011 7 890 0065 1 21 40 105-6691 1 1 1 731 3194 3 4 1 338 8570 255 346 8524 1836 1 E-mail: {fukumoto,kawasaki}@ibe.kagoshima-u.ac.jp, ryo-f@hiroshima-cu.ac.jp, fukuda@cv.ics.saitama-u.ac.jp, ymgc-tkm@signal.co.jp RGB transmission map Dehazing, transmission map,,,, 1. He Dehazing [3] Dehazing 2. Dehazing [1] [8] Robby He Dark Channel Prior [3] [1] Fattal 2. 1 Dehazing [2] He 1 [3] Dehazing [4], [5] I(x) = J(x)t(x) + A(1 t(x)) (1) 2 I J A t (1) J J I t A IS3-37 : 1111
(4) ω t(x) = 1 ω min Ω ( (I C (y))) min 0 < ω < C A C = 1 (5) ω (5) t transmission map tmap 1 4(a) 2. 3 2. 2 t 4(a) t tmap RGB 2 (a) RGB (A), (B), (C) I i = α i F i + (1 α i )B i (6) 2 (a) RGB 2 (b) F B α RGB (6) 2 (c) 3(a) α i ai i + b i w i (7) 3(b) a = 1 F B b = B F B w J(α) = min J(α, a, b) (8) a,b RGB α a, b ( min I C (x) ) ( = t(x) min J C (x) ) + C {r,g,b} C {r,g,b} ( 1 t(x) ) A C J(α) = α T Lα (9) (2) L Matting Laplacian C RGB α Levin Matting Laplacian [9] (5) tmap t(x) ( min min y Ω(x) C {r,g,b} IC (y) ) tmap t(x) t, t = t(x) ( ( J C (y) )) + ( 1 t(x) ) A C (3) E(t) = t T Lt + λ(t t) T (t t) (10) min y Ω(x) min C {r,g,b} Ω y t 1 [3] 2 λ t t(x) = 1 min Ω ( (I C (y))) min C A C (4) (L + λu)t = λ t (11) min Ω (min C ( IC (y) )) 1 t(x) 0 A C U L λ 1.0 10 4 tmap 4(b) IS3-37 : 1112
(a) (b) RGB (c) 2 RGB (a) (b) 3 (a) transmission map (tmap) 4 (b) tmap tmap 2. 4 A A A 2. 5 Dehazing t A (1) J (1) J(x) = I(x) A max(t(x), t 0 ) + A (12) t(x) 0 IS3-37 : 1113
(a) 3(a) (b) Dehazing 5 Dehazing Dehazing 3. 3 tmap t 0 = 0.1 t 0.1 5(b) Dehazing 3. 3. 1 tmap tmap Dehazing J 1) tmap ID 3. 4 tmap tmap 1 Dehazing 2 3 tmap tmap 4 tmap 5 Dark Channel Prior (1) Dehazing Dehazing tmap 3. 2 2 1) tmap ID 2) ID Dehazing ID 1 1) Dehazing 3. 5 2) Dehazing ID [10] [12] 1 tmap IS3-37 : 1114
5. tmap ID 4. 4. 1 tmap Antari Z1200 II 6 6 Dehazing tmap 15 15 3 3 ω = 0.95 λ = 1.0 10 4 SCOPE ICT t 0 = 0.1 ε = 1.5 10 5 101710002 21200002 4. 2 (LR030) (1) tmap 7 8 Dehazing [1] Robby T. Tan, Visibility in Bad Weather from a Single Image, Computer Vision and Pattern Recogni- tion, 2008. CVPR 2008. IEEE Conference on, 2008. (2) 9(a) [2] Raanan Fattal, Single Image Dehazing, ACM SIG- GRAPH 2008 papers, pp.72:1 72:9, 2008. [3] Kaiming He, Jian Sun, Xiaoou Tang, Single Image ID 9(b) Haze Removal Using Dark Channel Prior, Computer 9(a) Vision and Pattern Recognition, 2009. CVPR 2009. IEEE Conference on, pp.1956 1963, 2009. 9(b) [4] Y.Y. Schechner, S.G. Narasimhan and S.K. Nayar, Instant Dehazing of Images using Polarization, IEEE Conference on Computer Vision and Pattern (3) Recognition (CVPR), Vol.I, pp.325 332, Dec, 2001. 10 10(a) [5] Y.Y. Schechner, S.G. Narasimhan and S.K. Nayar, ID Polarization-Based Vision through Haze, Vol.42, No.3, pp.511 525, Jan, 2003. 10(b) [6] Peter Carr, Richard Hartley, Improved Single Image Dehazing Using Geometry, Digital Image Computing: Techniques and Applications, pp.103 110, 2009. (4) 11 11(a),(b) [7] S.G. Narasimhan and S.K. Nayar, Vision and the Atmosphere, International Journal on Computer Vision, Vol.48, No.3, pp.233 254, Jul, 2002. 6 2 12(a) ID 12(b) [8] Yong Du, Guindon, B. and Cihlar, J., Haze detection and removal in high resolution satellite image with wavelet analysis, IEEE Transactions on Geoscience and Remote Sensing, 40, 1, pp.210 217, 2002. [9] Anat Levin, Dani Lischinski, Yair Weiss, A Closed- (5) 13 Form Solution to Natural Image Matting, IEEE 13 (a) Transactions on Pattern Analysis and Machine Intelligence, pp. 228 242, February, 2008. 13 (b) [10], 2007- CVIM-158-26, pp.193 204, 2007. (6) 14 [11] Yuri Boykov, Olga Veksler, Ramin Zabih, Fast Approximate Energy Minimization via Graph Cuts, Dark Channel Prior IEEE Transactions on Pattern Analysis and Machine Intelligence, pp.1222 1239, November, 2001. [12] C. Rother, V. Kolmogorov, and A. Blake, grabcut : interactive foreground extraction using iterated graph cuts, ACM Trans. Graph., Vol.23(3), pp.309?-314, 2004. IS3-37 : 1115
6 7 tmap 8 Dehazing IS3-37 : 1116
(a) ID (b) 9 tmap (a) ID (b) 10 (a) 1 (b) 2 11 2 IS3-37 : 1117
(a) ID (b) 12 2 (a) 1 (b) 2 13 (a) (b) 14 IS3-37 : 1118