Abstract
We address the question of how to choose between different likelihood functions for motion estimation. To this end, we formulate motion estimation as a problem of Bayesian inference and compare the likelihood functions generated by various models for image formation. In contrast to alternative approaches which focus on noise in the measurement process, we propose to introduce noise on the level of the velocity, thus allowing it to vary around a given model. We show that this approach generates additional normalizations not present in previous likelihood functions. We numerically evaluate the proposed likelihood in a variational framework for segmenting the image plane into domains of piecewise constant motion. The evolution of the motion discontinuity set is implemented using the level set framework.
Supp: ONR N00014-02-1-0720, NSF SBR-9631682,0240148, NIH R01EY013875-01A1.
This is a preview of subscription content, log in via an institution.
Buying options
Tax calculation will be finalised at checkout
Purchases are for personal use only
Learn about institutional subscriptionsPreview
Unable to display preview. Download preview PDF.
References
Black, M.J., Anandan, P.: The robust estimation of multiple motions: Parametric and piecewise–smooth flow fields. Comp. Vis. Graph. Image Proc.: IU 63(1), 75–104 (1996)
Cremers, D.: A variational framework for image segmentation combining motion estimation and shape regularization. In: Dyer, C., Perona, P. (eds.) IEEE Int. Conf. on Comp. Vis. and Patt. Recog., Madison, Wisconsin (June 2003) (to appear)
Cremers, D., Schnörr, C.: Motion Competition: Variational integration of motion segmentation and shape regularization. In: Van Gool, L. (ed.) DAGM 2002. LNCS, vol. 2449, pp. 472–480. Springer, Heidelberg (2002)
Cremers, D., Soatto, S.: Combining motion estimation and surface regularization for variational image sequence segmentation. In: ICCV, Nice (October 2003) (to appear)
Horn, B.K.P., Schunck, B.G.: Determining optical flow. A.I. 17, 185–203 (1981)
Kohlberger, T., Memin, E., Schnörr, C.: Variational dense motion estimation using the Helmholtz decomposition. In: Griffin, L. (ed.) Int. Conf. on Scale Space Theories in Comp. Vis. LNCS, Springer, Heidelberg (2003) (to appear)
Kornprobst, P., Deriche, R., Aubert, G.: Image sequence analysis via partial differential equations. J. Math. Im. Vis. 11(1), 5–26 (1999)
Nagel, H.H.: Optical flow estimation and the interaction between measurement errors at adjacent pixel positions. Int. J. of Comp. Vis. 15(3), 271–288 (1995)
Nagel, H.H., Enkelmann, W.: An investigation of smoothness constraints for the estimation of displacement vector fields from image sequences. IEEE Trans. on Patt. Anal. and Mach. Intell. 8(5), 565–593 (1986)
Nestares, O., Fleet, D.J., Heeger, D.J.: Likelihoo functions and confidence bounds for total-least-squares problems. In: Proc. Conf. Computer Vis. and Pattern Recog., Hilton Head Island, SC, June 2000, vol. 1, pp. 760–767 (2000)
Osher, S.J., Sethian, J.A.: Fronts propagation with curvature dependent speed: Algorithms based on Hamilton–Jacobi formulations. J. of Comp. Phys. 79, 12–49 (1988)
Schnörr, C.: Computation of discontinuous optical flow by domain decomposition and shape optimization. Int. J. of Comp. Vis. 8(2), 153–165 (1992)
Simoncelli, E.P.: Distributed Representation and Analysis of Visual Motion. PhD thesis, Dept. of Elect. Eng. and Comp. Sci., MIT, Cambridge (1993)
Simoncelli, E.P., Adelson, E.H., Heeger, D.J.: Probability distributions of optical flow. In: Proc. Conf. Comp. Vis. Patt. Rec., Hawaii, June 1991, pp. 310–315 (1991)
Weickert, J., Schnörr, C.: A theoretical framework for convex regularizers in PDE–based computation of image motion. IJCV 45(3), 245–264 (2001)
Weiss, Y., Fleet, D.J.: Velocity likelihoods in biological and machine vision. In: Lewicki, M.S., Rao, R.P.N., Olshausen, B.A. (eds.) Probabilistic Models of the Brain: Perception and Neural Function, pp. 81–100. MIT Press, Cambridge (2001)
Author information
Authors and Affiliations
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2003 Springer-Verlag Berlin Heidelberg
About this paper
Cite this paper
Cremers, D., Yuille, A. (2003). A Generative Model Based Approach to Motion Segmentation. In: Michaelis, B., Krell, G. (eds) Pattern Recognition. DAGM 2003. Lecture Notes in Computer Science, vol 2781. Springer, Berlin, Heidelberg. https://doi.org/10.1007/978-3-540-45243-0_41
Download citation
DOI: https://doi.org/10.1007/978-3-540-45243-0_41
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-540-40861-1
Online ISBN: 978-3-540-45243-0
eBook Packages: Springer Book Archive