VR 1 1 1 1 1 SIFT Virtual View Generation in Hallway of Cybercity Buildings from Video Sequences Sachiyo Yoshida, 1 Masami Takata 1 and Joe Kaduki 1 Appearance of Three-dimensional (3D) building model is one of the most important components in a cyber city implementation and application. In this paper, we propose an approach for generating more freely and higher quality appearance of cyber city by using digital Video sequences than existing ones. This technique is updated depending on positions of the virtual view, and maps texture. In existing techniques, a large panoramic image is given by mosaicing all-directional camera.since the object of the techniques is large space such as city scenes, the techniques are unsuitable for small space such as indoor corridor. Therefore, the proposed a system aims to cyber city, which create automatically, for small space using homography, Scale Invariant Feature Transform (SIFT) and Video sequences. When users change their location in VR space, the system change textures newly. The system keeps realistic sensation and free from location s limits. 1. Google 5 4 1 1) 2) 3D 3D 360 1 Graduate School of Humanities and Sciences,Nara Women s Univercity 1 c 2010 Information Processing Society of Japan
2 3 4 2. VR 2.1 2.2 Google 2.1 3D 3 2.2 1 1 360 3. 3.1 3.2 3.3 3.4 3.5 3.6 3.7 3.8 GUI 3.1 2 2 3 3 1 3.2 2 c 2010 Information Processing Society of Japan
1 3 3 3.2 360 720 480 Bmp Microsoft Windows Bitmap Image 3.3 2 5 8 8 8 LK (Lucas Kanade 3) Scale-Invariant Feature Transform SIFT 4) LK SIFT SIFT SIFT 2 1 x h 11 h 12 h 13 x y = h 21 h 22 h 23 y (1) w h 31 h 32 h 33 w (x, y, w) (x, y, w ) h 11 h 33 x (h 31 x + h 32 y + h 33 ) = h 11 x + h 12 y + h 13 (2) y (h 31x + h 32y + h 33) = h 21x + h 22y + h 23 (3) 3 c 2010 Information Processing Society of Japan
(2) (3) 4 5 2 2 3.4 5 5 RANSAC RANdom Sample Consensus 5) 1 Boykov Interactive Graph Cuts 6) seed minmum/maximum flow algorithm 2 3 2 seed 2 seed 3 seed 4 3.5 3 2 4 SLAM Simultaneous Localization And Mapping 7) SLAM SLAM 3.6 3 OpenGL Open Graphics Library OpenGL 4 c 2010 Information Processing Society of Japan
5 GUI 6 2 4 3D 3.7 UI UI 3D 5 3 3.8 GUI 3.3 firstframe GUI 8 2 3 7 1 Probabilistic Hough Transform 5 2 4 4 8 4. victor GZ- M505-B OS windows XP 3 visualstudio2005 C Intel OpenCV Open Source Computer Vision Library OpenGL OpenCV 1 RANSAC 1000 1 3.1458 5 c 2010 Information Processing Society of Japan
8 (2) 9 (3) 2 6 7 8 9 OpenGL 7 7 7 2 6 8 9 5. GUI GUI GUI 3 3 1) Tsai, F., Chen, C.-H., Liu, J.-K. and Hsiao, K.-H.: Texture Generation and Mapping Using Video Sequences for 3D Building Models, Springer Berlin Heidelberg, pp.429 438 (2006). 2) Coorg, S. and Teller, S.: spherical mosaics with quaternions and dense correlation, nternational Journal of Computer Vision, Vol.37, pp.259 273 (2000). 3) Harris, C., Stephens, M. and Manor, R.: A combined corner and edge detector, Proceedings of The Fourth Alvey Vision Conference, pp.147 151 (1988). 4) : Gradient -SIFT HOG-, CVIM 160, pp.211 224 (2007). 5) Martin A. Fischler and Robert C. Bolles.: Random sample consensus: a paradigm for model fitting with applications to image analysis and automated cartography, Commun. ACM, Vol.24, number 6, pp.381 395 (1981). 6) Yuri Y. Boykov and Marie-Pierre Jolly.: nteractive Graph Cuts for Optimal Boundary & Region Segmentation of Objects in N-D Images, Eighth IEEE International Conference, Vol.1, pp.105 112 (2001) 7) Howie Choset and K. Nagatani: Topological simultaneous localization and mapping (SLAM): towardexact localization without explicit localization IEEE Transactions on Robotics and Automation, Vol.17, pp.125 137 (2001) 6 c 2010 Information Processing Society of Japan