Diffusion of Geometric Affinity for Surface Integration
A combinatorial method is used to reconstruct a surface by integrating a field of surface normals. An affinity function is defined over pairs of adjacent locations. This function is based on the surface’s principal curvature directions, which are intrinsic and can be estimated from the surface normals. The values of this locally supported function are propagated over the field of surface normals using a diffusion process. The surface normals are then regularised, by computing the weighted sum of the affinity evolved over time. Finally, the surface is reconstructed by integrating along integration paths that maximise the total affinity. Preliminary experimental results are shown for different degrees of evolution under the presence of noise.
KeywordsHeat Kernel Surface Normal Trapezium Rule Surface Integration Adjacent Location
- 3.Agrawal, A., Chellappa, R.: An algebraic approach to surface reconstruction from gradient fields. In: Proceedings of ICCV 2005, pp. 23–114 (2005)Google Scholar
- 4.Agrawal, A., Chellappa, R.: What is the range of surface reconstructions from a gradient field. In: Proceedings of ECCV 2006 (to appear, 2006)Google Scholar
- 6.Klette, R., Schlüns, K.: Height data from gradient fields. In: Proceedings of SPIE, vol. 2908, pp. 204–215 (1996)Google Scholar
- 7.Kondor, R.I., Lafferty, J.: Diffusion kernels on graphs and other discrete structures. In: ICML 2002 (2002)Google Scholar
- 8.Zhang, F., Qiu, H., Hancock, E.: Evolving spanning trees using the heat equation. In: Int. Conf. on Computer Analysis of Images and Patterns (2005)Google Scholar