Abstract
Multi-Grid methods for solving large linear or nonlinear equations, which are typical for early vision problems, are an important tool to solve these tasks in appropriate time. The standard method to estimate a displacement vector field from two images leads to a linear laplacian equation. Discontinuities can be included in this concept. A Multi-Grid approach can solve this equation with a fixed iteration number of a conjugate gradient smoothing operator under use of special design conditions. Convergence results show the possibility of a realtime implementation of this displacement vector field approach.
Preview
Unable to display preview. Download preview PDF.
References
Braess D.: On the Combination of the Multigrid Method and Conjugate Gradients. In: Hackbusch W., Trottenberg U.(Ed.) Multigrid Methods II, Springer Berlin-Heidelberg-New York 1986, 52–64
McCormick S.F.: An Algebraic Interpretation of Multigrid Methods. SIAM J. Numer. Anal. Vol 19, No. 3, (1982), 548–560
Dengler J.: Local Motion estimation with the Dynamic Pyramid. Proc. 8th ICPR, Paris 1986, pp. 1289–1292
Enkelmann W.: Investigations of Multigrid Algorithms for Estimation of Optical Flow Fields in Image Sequences. Proc. Workshop on Motion. Charl. S.C. 1986, pp. 81–87
Hackbusch W.: Multi-Grid Methods and Applications. Springer Berlin-Heidelberg-New York-Tokyo 1985.
Nagel H.-H.: Image Sequences — Ten (octal) Years From Phenomenology towards a Theoretical Foundation. Proc. 8th ICPR, Paris 1986, pp. 1174–1185
Schmidt M.: Mehrgitterverfahren zur 3D Rekonstruktion aus 2D Ansichten. Ph.D. Dissertation, University Heidelberg 1988
Terzopoulos D.: Multilevel Computational Processes for Visual Surface Reconstruction. CVGIP 24, (1983), 52–96
Author information
Authors and Affiliations
Editor information
Rights and permissions
Copyright information
© 1989 Springer-Verlag Berlin Heidelberg
About this chapter
Cite this chapter
Schmidt, M., Dengler, J. (1989). Adapting multi-grid-methods to the class of elliptic partial differential equation appearing in the estimation of displacement vector fields. In: Cantoni, V., Creutzburg, R., Levialdi, S., Wolf, G. (eds) Recent Issues in Pattern Analysis and Recognition. Lecture Notes in Computer Science, vol 399. Springer, Berlin, Heidelberg. https://doi.org/10.1007/3-540-51815-0_60
Download citation
DOI: https://doi.org/10.1007/3-540-51815-0_60
Published:
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-540-51815-0
Online ISBN: 978-3-540-46815-8
eBook Packages: Springer Book Archive