作者: David Forsyth , Andrew Zisserman , Jitendra Malik
关键词:
摘要: Recognising 3D objects from single images presents a range of significant problems, mostly to do with the nature and distinctiveness representations that can be recovered. In such special cases as polyhedra, surfaces revolution, general cones, canal algebraic surfaces, geometry recovered varying degrees success ambiguity. We discuss these volumetric primitives, comparing their utility surface primitives.