Study of Data Partition Algorithm for Irregular Meshes

A novel partition method is introduced, in which the traditional edge detection approach and region growing approach are integrated together to achieve robust partition from point data with moderate noise. First of all, the edge detection process will be used to extract edge regions, which include mesh elements near edge curves and coarsely segment the mesh surface into disjoint patches. Then the edge regions will be thinned into edge polygons by simultaneously growing the mesh patches to absorb compatible vertices. Point clouds sampled from objects of complex shapes with different scanning devices will be experimentally tested. The partition results point out that the developed algorithm is reasonably robust to noise.