Méthodes de Reconstruction Denses pour la Vision Active
Dense Reconstruction Methods for Active Vision
OPEN ACCESS
This paper aims to analyse how to introduce 3D information into an active vision system . In order to do so, we propose to realize a dense reconstruction from a monocular sequence in an active vision application . We first present existing algorithms for dense reconstruction. After having compared their drawbacks and advantages, we describe the chosen algorithm . Finally, we show the results obtained from synthetic images and images acquired when using an artificial robotic head.
Résumé
Ce papier cherche à analyser comment introduire des données tridimensionnelles au sein d'un système de vision active. En effet, nous nous sommes proposés de réaliser une reconstruction dense 3D à partir de l'analyse d'une séquence monoculaire, dans un paradigme de vision active. Nous présentons tout d'abord les différents algorithmes déjà existants dans le domaine de la reconstruction dense, en faisant ressortir leurs avantages et leurs inconvénients . Nous décrivons, ensuite, l'algorithme choisi pour pallier à certains de ces inconvénients. Enfin, nous montrons quelques résultats à partir d'images synthétiques ou de vues réelles acquises par la tête artificielle.
Active vision, Fast 3D reconstruction, Region segmentation, Regularization
Mots clés
Vision active, Reconstruction dense en 3D rapide, Segmentation région, Régularisation
[1] T. Aach and A . Kaup, M .A .P. estimation of dense disparity-fields for stereoscopic images . In international Conference on Image Procesing, Singapore , pages 1113-117, September 1992 .
[2] A . L. Abbott and N . Ahuja, Surface reconstruction by dynamic integration of focus, camera vergence, and stereo . In International Conference on Computer Vision, pages 532-543, Tampa, FL, Decembre 1988 .
[3] A . L . Abbott and N. Ahuja, Active Surface reconstruction by integrating focus , vergence stereo and camera calibration. In Proceeding of the 3 rd ICCV, Osqka, pages 489-492, 1990.
[4] N . Ayache, Artificial Vision for Mobile Robots. MIT Press, Cambridge, Massachusetts, 1989 .
[5] L. Cohen, L. Vinet, P. Sander, and A . Gagalowicz, Hierarchical region based stereo matching. In CVPR' 89 San-Diego, pages 416-421, 1987 .
[6] P. Duchateau et D . W. Zachmann, Theory and problem of Partial Differential Equations . Schaum's oytline series, Mc Graw-Hill Book Compagny, 1986 .
[7] J . Fairfield, Toboggan contrast enhancement. in Application of Artiricial Intellingence, Machine Vision and Robotics, volume 1708, pages 221-229. Proceedings of S .P.I .E ., 1980 .
[8] O .D . Faugeras, Three-dimensional Computer Vision : a geometric viewpoint. MIT Press, Boston, 1993 .
[9] O .D . Faugeras, B . Hotz, H. Mathieu, T. Viéville, Z. Zhang, P. Fua, E. Théron , L . Moll, G . Berry, J . Vuillemin, P. Bertin, and C. Proy, Real time correlationbased stereo : algorithm, implementations and applications . Technical report 2013, INRIA, 1993 .
[10] W. Foerstner and A . Pertle, Photogrammetric standard methods and digital images matching techniques for high precision measurements. Patterm recognition in pratice, pages 55-72, 1986.
[11] P. Fua, A parallel stereo algorithm that produces dense depth maps and preserves images features . Technical Report 1369, Institut National de recharche en Informatique et en Automatique, January 1991 .
[12] D. Geman and G . reynolds, Constrained restoration and the recovery of discontinuities. Transactions on Pattern Analysis and Machine Intelligence, 14(3), March 1992 .
[13] W. Hoff and N . Ahuja, Extracting surfaces from stereo images : An integrated approach. in 1st ICCV, London, pages 284-294, 1987.
[14] B K P Horn, Robot Vision . MIT Press, Cambrodge, Massachusetts, 1986 .
[15] B K P Horn, Height and gradient from shading . International journal of Computer Vision, 5 : 1 : 37- 75, 1990.
[16]J .J . Hwang and E.L . Hall . Matching of featured objects using relational tables form stereo images . Computer Graphics and Image Processing, 20 : 22-42, 1982.
[17] S . Mitter, J . Marroquin and T. Poggio, probabilistic solution of ill-posed problems in computation vision . Journal of the American Statistical Association , 82(397) : 76-89, March 1987 .
[18] T. Kanade and M. Okutomi, A stereo matching algorithm with an adaptative window : Theory and experiment . Artificial intelligence CMU-CS-90-120 , Carnegie Mellon, April 1990 .
[19] R . Deriche , L. Robert and O . D. Faugeras, Dense Depth Map Reconstruction Using Multiscale Regularization . In 2 nd Singapore International Conference on Image Processing, pages 13-127, Singapore , September 1992.
[20] H . Maître and W. Luo, Using models to improve stereo reconstruction. IEEE trans. on Pattern Analysis and Machine Intelligence, 14: 269-277, 1992.
[21] R . March, A regularization model for stereo vision with controlled continuity . Pattern Recognution Letters, pages 259-263, october 1989 .
[22] J.L. Marroquin, Surface reconstruction preserving discontinuities . Artificial Intelligence A. I. MEMO 792, Massachusetts Institute of Technology, August 1984.
[23] K. Pahlavan, J .O. Ekhlund, and T. Uhlin, Dynamic fixation . In 4th ICCV, Berlin, pages 404-411 . IEEE Society, 1993 .
[24] G. Randall, parallélisation d'un algorithme de stéréoscopie trinoculaire . PhD thesis, University of Orsay, Dept of Comp . Science, 1991 . PhD thesis .
[25] Y. Remion, H. Maître, and J .L. Krahe, Recalage pare zones de 2 vue stéréoscopiques d'un univers composé d'objets plans . In Cognitiva'87, La Villette, Paris, pages 175-180, 1987.
[26] L. Robert, Stéréovision : de la mise en correspondence de courbes à l'analyse photogrammatrique de la scène . PhD thesis, Ecole polytechnique, Dept of Comp. Science, 1993 . PhD thesis .
[27] P.S . Toh and A.K. Forrest, Occlusion detection in early vision . in Third International Conference on Computer Vision, Osaka, pages 126-132, December 1990.
[28] R . Vaillant and O.D . Faugeras, Using extremal bounderies for 3d object modeling . In IEEE transaction on >pattern Analysis and Machine Intelligence, volume 14, pages 157-173, February 1992.
[29] A . Verri and T. Poggio, Motion field and optical flow : differences and qualitative properties. Technical report AI Memo 917, MIT Press, Cambridge , 1986
[30] T. Viéville, E. Clergue, R . Enciso, and H . Mathieu, Experimentating with 3d vision on a robotic head . Robotics and Autonomous Systems, 14(1), 1995 .
[31] T. Viéville and Q .T. Luong, Computing motion and structure in image sequences without calibration . In The 12th Int. Cof on Pattern Recognition , Jerusalem, pages 420-426, 1994.
[32] A . Witkin, D . Terzopoulos, and M . Kass, Signal matching trough scale space. International journal of Computer Vision, pages 133-144, 1987 .
[33] B . Wrobel-Dautcourt, Surfaces trìdimentionnelles obtenues par appariement stéréoscopique de régions . in Sixième AFCET en reconnaissance des formes et intelligence artificielle, Paris, pages 299-307, 1987 .
[34] N. Yokoya, Stereo surface reconstruction by multiscale-multistage regularization. Technical report 90-45, Electrotechnical Laboratory, November 1990 .
[35] Z. Zhang, R. Deriche, Q-T. Luong, and O . Faugeras, A robust approach to image matching : Recovrey of the epipolar geometry. In Proc. International Symposium of Young Investigators on Information/Computer/Control, pages 7-28, Beijing, China, February 1994.