Presents a method to compute the inter-frame transformationbetween two range image views of complex multi-part objects. No exactfeature matching is attempted and no initial approximate transformationis provided. The method is naturally decomposed into two stages ofinitial estimation and final refinement of the transformation. Ahierarchical triangulation-based surface representation provides anefficient way to select the control points at which the alignment of thetwo surfaces is to be evaluated. This representation also permits theselection of a manageable number of initial transformations among whichat least one is to be in the parametric neighborhood of the actualtransformation. Experimental results show that the computedtransformation between two views of a complex multi-part object mayprovide angles of rotation within a fraction of a degree of the actualones
展开▼