Robust Multiresolution Estimation of Parametric Motion Models in Complex Image Sequences
Estimation Robuste Multiéchelle de Modèles Paramétrés de Mouvement sur des Scènes Complexes
OPEN ACCESS
This paper describes a parametric motion model estimation algorithm. Motivations for the use of such models are on the one hand their compactness, which has proved to be efficient in numerous contexts such as estimation, segmentation, tracking and interpretation of motion, and on the other hand, the low computational cost of its estimations. However, it is important to have the best accuracy for the estimated parameters, and to take into account the problem of multiple motion. We therefore developed two M-estimator-like estimators in a multiresolution framework. Numerical results support this approach, as demonstrated by the use of these algorithms on complex sequences.
Résumé
Nous présentons dans cet article une méthode d'estimation de modèles paramétrés de mouvement. L'intérêt de formuler un problème d'analyse de mouvement par l'identification de tels modèles est double. Il s'agit d'une part d'une représentation compacte qui s'avère adéquate et pertinente dans des contextes différents et nombreux (mesure, segmentation, suivi ou caractérisation du mouvement). D'autre part, une estimation peu coûteuse en temps calcul peut en être obtenue. Le point crucial par contre est d'en établir une estimation fiable et précise, et de gérer correctement la présence de plusieurs mouvements dans l'image . Pour répondre de façon bien fondée et efficace à ces problèmes, nous proposons deux versions d'un estimateur robuste, du type M-estimateur, défini dans un schéma multirésolution . Cette technique a permis d'obtenir des résultats tout-a-fait satisfaisants sur des images représentant des scènes complexes.
Dynamic scene analysis, parametric models, motion estimation, robust estimation, multiresolution, segmentation.
Mots clés
Séquence d'images, mouvement, modèles paramétrés, estimateur robuste, mesure multiéchelle, segmentation.
[1] J.K. Aggarwal and N. Nandhakumar. On the computation of motion from sequences of images- a review. Proc. of the IEEE, Vol.76, No.8 :917-935, August 1988.
[2] B. Bascle and N. Deriche. Stereo matching, reconstruction and refinement of 3D curves using deformable contours. InProc. 4th Int. Conf Computer Vision, Berlin,pages 421-430, May 1993.
[3] R. Battiti, E. Amaldi, and C. Koch. Computing optical flow across multiple scales : an adaptative coarse-to-fine strategy. Intern. J. Comput. Vis., 6:2:133-145,1991.
[4] J.R. Bergen, P. Anandan, K. Hanna, and R. Hingorani. Hierarchical modelbased motion estimation. InProc. of 2nd European Conf. Computer Vision, pages 237-252, Springer-Verlag, S.Margherita Ligure, Italic, 1992.
[5] J.R. Bergen, P.J. Burt, R. Hingorani, and S. Peleg. Computing two motions from three frames. InProc. 3rd Int. Conf on Computer Vision, pages 27-32, Osaka, Déc. 1990.
[6] M. J. Black. Robust incremental optical flow. PhD thesis, N° 923, Yale University, Computer Science Dept, Septembre 1992.
[7] P. Bouthemy. Modèles et méthodes pour l'analyse du mouvement dans une séquence d'images. Technique et Science Informatiques,Vol.7, No.6:527546, 1988.
[8] P. Bouthemy and E. François. Motion segmentation and qualitative dynamic scene analysis from an image sequence. Int. Journal of Computer Vision, Vol.10, No 2 :157-182, Avril 1993.
[9] P.J. Burt. The pyramid as a structure for efficient computation. In A. Rosenfeld, editor, Multiresolution Image Processing and Analysis, pages 6-35, Springer-Verlag, 1984.
[10] P.J. Burt, R. Hingorani, and R.J. Kolczynski. Mechanisms for isolating component patterns in the sequential analysis of multiple motion . In IEEE Workshop on Visual Motion, pages 187-193, Princeton,Octobre 1991.
[11] T. Darrell and A. Pentland. Robust estimation of a multi-layered motion representation. InProc. IEEE Workshop on Visual Motion, pages 173-178, Princeton, Oct. 1991.
[12] W.Enkelmann. Investigations of multigrid algorithms for the estimation of optical flow fields in image sequences.Computer Vision, Graphics and Image Processing, Vol.43 :150-177, 1988.
[13] F.R. Hampel, E.M. Ronchetti, P.J. Rousseeuw, and W.A. Stahel. Robust Statistics : The Approach Based on Influence Functions. John Wiley and Sons, New York, 1986.
[14] M. Hoetter. Differential estimation of the global motion parameters zoom and pan. Signal Processing, Vol.16 :249-265, 1989.
[15] P.W. Holland and R.E.Welsch.Robust regression using iteratively reweighted least squares.Commun. Stat.- Theor. Meth.,A6:813-828, 1977.
[16] B.K.P. Horn and B.G. Schunck. Determining optical flow.Artificial Intelligence,Vol. 17:185-203, 1981.
[17] P.J. Hubert. Robust statistics. Wiley, 1981.
[18] M. Irani, B. Rousso, and S. Peleg. Detecting and tracking multiple moving objects using temporal integration. InProc. of 2nd ECCV-92, S.Margherita Ligure,Italie,pages 282-287, Springer-Verlag, Mai 1992.
[19] J.M. Jolion, P.Meer, and S. Batauche. Robust clustering with application in computer vision. IEEE Trans. Pattern Analysis and Machine Intelligence, 13 :791-802, Août1991.
[20] J. Konrad and E. Dubois. Multigrid Bayesian estimation of image motion fields using stochastic relaxation. InProc. 2nd Int. Conf. Computer Vision, pages 354-362, Tarpon Springs, Floride, Déc. 1988.
[21] G.E. Mailloux, F. Langlois, P.L. Simard, and M. Bertrand. Restoration of the velocity field of the heart from two-dimensional echocardiograms . IEEE Trans. on Medical Imaging, 11(2):143-153, Juin 1989.
[22] P.Meer,D. Mintz, and A.Rosenfeld.Robust regression methods for computer vision : a review. International Journal of Computer Vision, 6(1) :59-70, 1991.
[23] F. Meyer and P. Bouthemy. Estimation of time-to-collision maps from first order motion models and normal flows. InProc.11th intern. Conf. on Pattern Recognition, LaHaye,pages 78-82, 1992. [24] F.G. Meyer and P. Bouthemy. Region-based tracking using affine motion models in long image sequences. CVGIP : Image Understanding, Vol. 60(2) :119-140,Septembre 1994.
[25] H.H. Nagel. From image sequences towards conceptual descriptions. Image and Vision Computing Jal, Vol.6, No.2 :pp 59-74, Mai 1988.
[26] S. Negahdaripour and S. Lee. Motion recovery from image sequences using first-order optical flow information. InProc.of the IEEE Workshop on Visual Motion, Princeton, pages 132-139, Oct. 1991.
[27] H. Nicolas and C. Labit. Global motion identification for image sequence analysis and coding. In Proc. Int. Conf. on Acoustics, Speech, and Signal Processing, pages 2825-2828, Toronto, Mai 1991.
[28] P.J. Rousseeuw. Least median of squares regression. Jal of the American Statistical Association, Vol.79, No 388 :871-880, Déc. 1984.
[29] J. Schmetz and M.S. Mhita. Diurnal and interdiurnal variability ofIRand W V brightness temperatures from Meteosat. ESA Journal, 13 :329-341, 1989.
[30] T. Viéville and O. Faugeras. Robust and fast computation of unbiased intensity derivatives in images. InProc. of 2nd ECCV-92, S.Margherita Ligure, Italic, pages 203-212, Mai 1992.
[31] S. F. Wu and J. Kittler. A gradient-based method for general motion estimation and segmentation. Jal of Visual Communication and Image Representation, 4(l) :25-38, Mars 1993.
[32] W.-Z. Zhao, F.-H.Qi, and T.Y. Young. Dynamic estimation of optical flow field using objective functions. Image and Vision Computing, Vol.7, No.4 :259267, Nov. 1989.