Mesh Segmentation and Model Extraction

摘要: High precision laser scanners deliver virtual surfaces of industrial objects whose accuracy must be evaluated. But this requires the automatic detection reliable components such as facets, cylindric and spherical parts, etc. The method described here finds automatically parts in surface to which geometric primitives can fitted. Knowing certain properties input object, primitive fitting helps quantifying an acquisition process scanned mires. combines mesh segmentation with model fitting. is based on level set tree a scalar function defined mesh. applied simplest available intrinsic mesh, mean curvature. In first stage fast algorithm extracts sets function. Adapting meshes well known for extracting Maximally Stable Extremal Regions from digital images, segments into smooth separated by high curvature regions (the edges). This followed selection each part permitting fit planes, cylinders spheres quantify overall process.