slides - DIMAp
Transcription
slides - DIMAp
Marcelo Ferreira Siqueira UFMS - Brazil Joint work with Jean Gallier Dimas Morera Luis Gustavo Nonato CIS - UPenn - USA ICMC - USP - Brazil ICMC - USP - Brazil Dianna Xu Luiz Velho CS - Bryn Mawr - USA IMPA - Brazil Problem Statement Given a simplicial surface, ST , in R3 , with an empty boundary, a positive integer k, and a positive real number , we want to ... ST 2 Problem Statement find a C k surface, S, in R3 such that . . . S ⊂ R3 3 Problem Statement there exists a homeomorphism, h : |ST | → S, satisfying h(v) − v ≤ for every vertex v in ST . |ST | S 4 Problem Statement REMARK: ST is expected to be “very large” (∼ 106 vertices). 5 An Adaptive Fitting Approach Step 1: Simplify ST using the Four-Face Clusters algorithm. ST ST See [Velho, 2001] 6 An Adaptive Fitting Approach Algorithm preserves topology. Each vertex of ST is a vertex of ST . ST ST ST is also a hierarchical multiresolution mesh. 7 An Adaptive Fitting Approach Step 2: Map the edges of ST to ST using geodesics. 8 An Adaptive Fitting Approach Step 2: (continuation...) Re-triangulate ST so that the geodesics are covered by edges. Adapted from the algorithm in [Morera, Carvalho and Velho, 2005] 9 An Adaptive Fitting Approach Step 3: Parametrize the star of each vertex v of ST over a regular polygon inscribed in a unit circle in R2 and containing the vertex (0,1). v 10 An Adaptive Fitting Approach Step 3: (continuation...) Map the vertices of ST to the regular polygons. v We use Floater’s parametrization for each “macro triangle”. 11 An Adaptive Fitting Approach Step 4: ∞ , we define a C function, For each vertex v ∈ S Non-polynomial convex combination of Bézier patches! T γv : R2 → R3 through a least squares fitting using the parameter points in the polygon associated with v and their corresponding vertices in ST . ST w w 12 An Adaptive Fitting Approach Step 4: (continuation...) Compute the approximation error: γv (w ) − w. If γv (w ) − w ≥ then ST must be locally refined. Refinement is simple: we take advantage of the hierarchical and multiresolusion structure of ST . This comes from the simplification algorithm. After all faces are refined, we go back to Step 2. 13 An Adaptive Fitting Approach Step 5: We define a parametric pseudo-manifold, M, in R3 using the topology of ST , the vertices of ST , and the parametrizations computed in Step 3. M ST M is the image of M in R3 . 14 Gluing Data and PPM’s Rm θ1 θ2 Rn ϕ12 Ω1 Ω21 Ω12 ϕ21 See [Grimm and Hughes, 1995] 15 Ω2 Gluing Data and PPM’s Rm θ1 θ2 θi (p) Rn θj ◦ ϕ21 (p) ϕ12 Ω1 p Ω21 Ω12 ϕ21 16 Ω2 Gluing Data and PPM’s See [Siqueira, Xu, and Gallier, 2008] 17 Concluding Remarks The overall idea (mesh simplification + mesh parametrization) of the previous adaptive fitting is not new, but the components (i.e., geodesics and parametric pseudo-manifolds ) used in our solution make it simpler and/or more powerful than similar approaches. The work is still in progress... Code for computing geodesics and re-triangulate ST is not stable. 18 Concluding Remarks Code for computing parametric pseudo-surfaces is finished. 19 References • L. Velho. Mesh Simplification Using Four-Face Clusters. In Proceedings of the International Conference on Shape Modeling & Applications (SMI), 2001. • D. Morera, L. Velho, and P.C. Carvalho. Computing Geodesics on Triangular Meshes, Computer & Graphics, 29(5): 667-675, 2005. • C. M. Grimm and J. F. Hughes. Modeling Surfaces of Arbitrary Topology Using Manifolds. In Proceedings of the ACM SIGGRAPH, 1995. • M. Siqueira, D. Xu, and J. Gallier. Construction of C ∞ Surfaces from Triangular Meshes Using Parametric Pseudo-Manifolds. Technical Report MS-CIS-08-14, Department of Computer and Information Science, University of Pennsylvania, 2008. 20