Image reconstruction from point cloud data by CIP-level set method

Level set method(LSM) is the efficient computation method for the interface capturing with corresponding of advection equation attend topological change of interface. LSM is consisted of the generation of signed distance function from initial interface and the convergence computation of the interface evolution for the target shape using the advection equation in conjunction with the mean curvature flow. We can get the image source from the actual object through the some kinds of distance measurement. The many measured points form the point cloud as the pointing sketch of a object surface. The precision surface model of the object requires that we extract the suitable interface for the purpose from the point cloud data. The extraction is one of image reconstruction. LSM is very useful method to extract the fine surface model from the point cloud data. However, the conventional LSM has some difficalties in the practical computing. In the long time computing, the distance function will be broken up without the frequent reinitialization of it. In this study, we proposed the new computation algorithm in which it is combined the LSM with the CIP scheme. The CIP presents very good performance to keep the shape profile in the advection computing. The obtained results are the various of examples of image reconstruction such as the convex sharp profile, the non-convex profile and the topological changing of profiles. We show the applicability and the effectiveness through the comparison of the CIP scheme with Up-Wind scheme.