PASCAL - Pattern Analysis, Statistical Modelling and Computational Learning

Optimal Ratio of Lamé Moduli with Application to Motion of Jupiter Storms
Ramunas Girdziusas and Jorma Laaksonen
In: 14th Scandinavian Conference on Image Analysis, 19-22 Jun 2005, Joensuu, Finland.

Abstract

Fluid dynamics models the distribution of sources, sinks and vortices in imaged motion. A variety of different flow types can be obtained by specifying a key quantity known as the ratio of the Lamé moduli λ/μ. Special cases include the weakly elliptic flow λ/μ→ –2, often utilized in the Monge-Ampère transport, the Laplacian diffusion model λ/μ =–1, and the hyper-elliptic flow λ/μ → ∞ of the Stokesian dynamics. Bayesian Gaussian process generalization of the fluid displacement estimation indicates that in the absence of the specific knowledge about the ratio of the Lamé moduli, it is better to temporally balance between the rotational and divergent motion. At each time instant the Lamé moduli should minimize the difference between the fluid displacement increment and the negative gradient of the image mismatch measure while keeping the flow as incompressible as possible. An experiment presented in this paper with the interpolation of the photographed motion of Jupiter storms supports the result.

EPrint Type:Conference or Workshop Item (Paper)
Project Keyword:Project Keyword UNSPECIFIED
Subjects:Computational, Information-Theoretic Learning with Statistics
Machine Vision
Learning/Statistics & Optimisation
Theory & Algorithms
ID Code:1739
Deposited By:Jorma Laaksonen
Deposited On:28 November 2005