May 13 2015, Byron blanc.
In this presentation we first introduce an iterative projection technique for parameterizing and fitting exactly a point cloud as a sample of a 4-edges (data) surface. This method is mainly based on the estimation of the so called “closest point parameters”, which requires an initial fitting surface computed from its four edges B-spline fitting curves. We then present a simple baycentric alternative technique to compute the parameters of this initial surface. In a third part, as a generic improvement tool for surface fitting with B-splines, we introduce a lapacian correction matrix applied to the control points network. We will finally review some examples to demonstrate the benefit on the fitting surface accuracy, before concluding by a discussion of the improvements provided in this way, regarding to the state of the art.
Keywords : surface fitting, B-spline surface, point cloud, initial surface, Coons patch, barycentric parameterization, closest point, edges, mesh, error optimization.
Contributions : barycentric patch, laplacian matrix for control points net, closest point parameters estimation.
This is a common seminar between GALAAD2 and TITANE teams.
