Mathematical morphology based surface reconstruction from defective and scan misregistration point sets

It is a difficult but important challenge to handle the defective and scan misregistration (DSM) parts in point sets for surface reconstruction. This paper presents a robust and convenient algorithm to overcome the problem. This method extends the operations of mathematical morphology in grey image process to reduce the DSM parts and confirms the oriented information of point sets. Based on the information, the Poisson equation is adopted to construct a watertight and smooth resulting surface. By this method, an implicit surface with global accuracy and smoothness can be obtained. This method can also give resilience to noise, adjustable fitting to the sparse data. Finally, a comparison between this method and some common methods is given to demonstrate its effect.