2010 | OriginalPaper | Chapter
High Quality Visual Hull Reconstruction by Delaunay Refinement
Authors : Xin Liu, Marina L. Gavrilova
Published in: Transactions on Computational Science IX
Publisher: Springer Berlin Heidelberg
Activate our intelligent search to find suitable subject content or patents.
Select sections of text to find matching patents with Artificial Intelligence. powered by
Select sections of text to find additional relevant content using AI-assisted search. powered by
In this paper, we employ Delaunay triangulation techniques to reconstruct high quality visual hulls. From a set of calibrated images, the algorithm first computes a sparse set of initial points with a dandelion model and builds a Delaunay triangulation restricted to the visual hull surface. It then iteratively refines the triangulation by inserting new sampling points, which are the intersections between the visual hull surface and the Voronoi edges dual to the triangulation’s facets, until certain criteria are satisfied. The intersections are computed by cutting line segments with the visual hull, which is then converted to the problem of intersecting a line segment with polygonal contours in 2D. A barrel-grid structure is developed to quickly pick out possibly intersecting contour segments and thus accelerate the process of intersecting in 2D. Our algorithm is robust, fast, fully adaptive, and it produces precise and smooth mesh models composed of well-shaped triangles.