Surface detection and modeling of an arbitrary point cloud from 3D sketching

ABSTRACTThis work describes a method for converting unorganized point cloud data from 3D sketching into an explicit 3D model, contributing to closing the knowledge gap in transferring 3D sketches from a conceptual design to the detailed design. The main challenge in the work is processing an unorganized point cloud where the points fill the volume of the object and do not only lie on the surface, as they would in most scanned data. The scope of the work is a single object, genus 0. The 3D model is constructed by building a loft feature through curves generated by slicing the point cloud. Firstly, the point cloud is sliced and points worked onto slicing planes; secondly, the external points that describe the contour of each slice are isolated using alpha shapes; lastly, smooth curves are created through the external points, and a loft is generated.