( ), 42, 5,2006 9 Acta Scientiarum Naturalium Universitatis Pekinensis, Vol. 42, No. 5 (Sept. 2006) 1) 2D23D 2),3) 3) ( 2),,310035 ; 3),,,100871) 2 :,,, ; ; ; ; TP 391141 Seamless 2D23D Texture Mapping LI Xiaolan 2),3) ZHA Hongbin 3) ( 2) Computer Science and Technology Department, Zhejiang Industry and Commerce University, Hangzhou, 310035 ; 3) National Laboratory on Machine Perception, Peking University, Beijing, 100871) Abstract Wavelete decomposition was performed to classify the source images into two types : the edge2riched image and the smooth image. Based on the Human Visual System theory, different combination techniques, the image stitching and the image blending methods, were selected to deal with different source images. A new algorithm was proposed to find an optimized stitching path in the overlapped area of Partial Texture Maps. View dependent weight was added into the ordinary blending weight, which made the generated texture map reveal some geometric properties of the 3D model. The experimental results verify the validity of the methods. Key words wavelet block ; human visual system ; multiresolution blending ; blending weight ; image stitching 0 2, : (1) 1 ( ), ; (2) 2D23D ( (c) (e), ),, :,, 2D23D 2D23D,,,,,, : (1) ; (2),, 1) (60333010) 973 (2004CB318005) 674 : 2005211207 ; : 2006206215
5 : 2D23D [1 ] : [13 ], (UV ),, UVmesh, 2 : (1),,UVmesh, ; (2),, [2,3 ] 2 2D I 1 I 2, 2, I I 1 I 2, I 3D 3D 2 3D Fig. 2 Parametization of 3D model I 1 I 2, I i, UVPtexture i i = 1, 2,, num, num, (UVPtexture i [14 ] ) (feathering) [4,5 ], UVPtexture i, [6,7 ] 3, :, Su 1) 2 I ov 1 I ov 2, Levin (wavelete block) ; [8,9 ], 2) 2, T 1, I ov 1 I ov 2 ( 2D ), 2 ;, 3D 3), Rocchini [10,11 ], Pighin [12 ], 4 :,,,,, 1, 2, 3, 1, UVmesh 3 3 Fig. 3 The flow chart of image combination 675
( ) 42 4) T 2, S1 60 %,, 2 ;, 2. 2,, ;,,, [14 ] 2 2. 1,, I 1 I 2, I ov 1 I ov 2,,, e i, j = abs( I ov 1 ( i, j) - I ov 2 ( i, j) ), i = 1,, 2,, w ov, j = 1,2,, h ov, w ov h ov 3,, e, ( 4 ) 3 E : E i,1 = e i,1 E i, j = e i, j + min ( E i - 1, j, E i - 1, j, E i - 1, j+1 ),, ( 4 ) j = 2,, h ov (1), E, P ath = arg min ( E i ) i = 1,, w ov 5 1, ( 5 (a1) ) 5 (b2), ( UVPtexture ) : (1) e i, j, - 1 ; T len ; (2), (3) i - 1 E i, j ; (4) i 5 4 Fig. 4 The wavelet block - 1, i, len (Path i ) ; (5) len ( Path i ) > T len - 5, E i, j, [1 ], 2 : 5 2, S1 ;, S2, ( 5 (b1) ),, number{abs( F ( u, v) ) > T f ( u, v) WB k 2. 3 } T n, WB k S1 ;, WB k S2, T f T n, ;, 676
5 : 2D23D 1 Fig. 1 The results of different image combination methods 5 Fig. 5 The demonstration of image stitching along optimizing path 6 2 Fig. 6 Comparing the results acquired with multiresolution blending and optimizing path stitching,,,,, : (1), M ; (2), T len, Keypos ; (3) Keypos,, 677
( ) 42 Num in < T len - 5, 3 ( 7 (a) ),, s l s r, ( 7 ( b) ) w = s r - s l + 1 ( 7 (c) ),,, 7(d), ( UV Ptexture), ( 7 (e) ), 7 (f) ( ),, UV Ptexture ( 7 (g) ) 7 (h),, ( ), ( (d), ), W = W m W f W v, (2) ; ( (h) ), ( (d) ), (1) W m, ( i, j) M, W m ( i, j) = 1 ; W m ( i, j) = 0 ; (2) W f, W f ( i, j) = 1 - ( i - s l )Πw ; (3) W v, ( i, j, D k ) - min ( i, j, D k ) ( i, j) M W v ( i, j) = max ( i, j, D k ) - min ( i, j, D k ), (3) ; ( (h) ) ( i, j) M ( i, j) M 8 ( i, j, D k ) ( i, j) 8 (a) 8 (b) 5 i, j k D k (, ) i, j = N ( v 1 ) + N ( v 2 ) + N ( v 3 ), N ( v i ) v i 3 ),, 8 (c) 8 (d), 1,,, : (1), 3 ; ( (h) ), ( (f) ),, (2,3, 4, 6 ; (2),,,, ( 6 (c) ) ;,, 2 ( 6(d) ) 5 (a5), 31504 16, 7 678
5 : 2D23D 7 Fig. 7 The definition of weight functions and their effects for the texture map combinating 8 Fig. 8 The rendering results from different viewpoints [ 1 ] BLAKEMORE C, CAMPBELL F W. On the existence of neurones in the human visual system selectively sensitive to the orientation and size of retinal images [ J ]. Journal of Physiology, 1969, 203 : 2372260. [ 2 ] EFROS A A, FREEMAN W T. Image quilting for texture synthesis and transfer [ C]. Proc of SIGGRAPH 2001. Los Angeles : ACM, 2001 : 3412346. [ 3 ] DAVIS J. Mosaics of scenes with moving objects[ C]. Proc of CVPR. Santa Barbara : IEEE, 1998: 3542360. [ 4 ] SZELISKIR,SHUMHY. Creatingfullviewpanoramic 679
( ) 42 image mosaics and texture2mapped models [ C ]. Proc of SIGGRAPHπ97. Los Angeles :ACM, 1997 : 2512258. [ 5 ] UYTTENDAELE M, EDEN A, SZELISKI R. Eliminating ghosting and exposure artifacts in image mosaics[ C]. Proc of CVPR. Hawaii : IEEE, 2001: 5092516. [ 6 ] BURT P J, ADELSON E H. A multiresolution spline with application to image mosaics[j ]. ACM ToG, 1983, 2 (4) : 2172236. [ 7 ] HSU C T, WU J L. Multiresolution mosaic[j ]. IEEE Trans Comsumer Electron, 1996, 42 : 9812990. [ 8 ] SU M S, HWANG W L, CHENG K Y. Analysis on multiresolution mosaic images [ J ]. IEEE Trans. Image Processing, 2004, 13 (7) : 9522959. [ 9 ] LEVIN A, ZOMET A, PELEG S, et al. Seamless image stitching in the gradient domain [ C ]. Proc of ECCVπ04. Prague : Springer2Verlag LNCS, 2004 : 3772389. [ 10 ] ROCCHINI C, CIGNONI P, MONTANI C. Multiple textures stitching and blending on 3D objcts [ C ]. Proc of 10 th Eurographics Workshop on Rendering. Granada, Spain : Springer, 1999 : 1272138. [ 11 ] ROCCHINI C, CIGNONI P, MONTANI C, et al. Acquiring, stitching and blending diffuse appearance attributes on 3D models[j ]. The Visual Computer, 2002, 18 (3) : 1862204. [ 12 ] PIGHIN F, HECKER J, LISCHINSKI D, et al. Synthesizing realistic facial expression from photographs[ C]. Proc of SIGGRAPHπ98. Los Angeles : ACM, 1998. 75284. [ 13 ] YAMAUCHI H, LENSCH H P A, HABER J, et al. Textures revisted[j ]. Visual Comput, 2005, 21 (4) : 2172 241. [14 ] LI X L, ZHA H. B. Realistic human head modeling with multi2view hairstyle reconstruction[ C]. Proc of the 5 th Inter Conf on 32D Digital Imaging and Modeling. Ottawa,Canada : IEEE, 2005 : 4322438. 3 3 3 3 3 T, 2 (CT ) NY2ESO21 (NY2ESO21b), NY2ESO21b (CSFDA), 863 2MHC (tetramer) IFN2 GranzymeBΠperforin ( ELISPOT) T (CTL), 1Π3 NY2ESO21b CTL ; ELISA, 3018 % NY2ESO21 ; HLA2A2 + NY2ESO21 mrna +,, NY2ESO21b NY2ESO21b,,, LUDWIG,, NY2ESO21b,, ( 2006209207) 680