Topologically consistent partial surface reconstruction from range pictures

Abstract

Surface or shape reconstruction from 3D digitizations, range pictures play an important role as the sizes and accessibility of the sampled object become intractable -- Range pictures, however, present challenges regarding: (i) recovery of topological structure from geometrical information of a partial view: (ii) inclusion of several or self obstructing objects on the same picture, and (iii) conciliation of partial topological and geometrical information from the individual views into a main model -- Issue (i) and (ii) require introduction of data structures and algorithms able to consistently represent incompleteness and discontinuities in the surface -- Aspect (iii) demands the application of statistical methods to sort redundant e inconsistent information in the overlaps between the individual views -- In this investigation, tasks (i) and (ii) have been undertaken by designing and populating an extended Boundary Representation (B-rep), using equivalence relations that induce partitions on the data sets -- Task (iii) has been carried out by using data processing tools (DigitLAB) that filter, resample, and recover shape from planar digitizations, by applying formalisms of differential topology