Generation of 3D Human Models with Different Levels of Detail through Point-Based Simplification

We describe a method for generating 3D human models with different levels of detail based on scanned point cloud. A simple hierarchical and interactive simplification algorithm is investigated to create approximations of the point-based human model with various sampling densities. With the completion of sampling, a triangulation procedure is performed to produce final meshes easily based on the sector tree information. Parameters allow users to control the mesh quality in terms of regularity, uniformity and density. The modeling algorithm has been tested with examples.