We present a method to compute the interframe transformation between t
wo range image views of complex multipart objects. No exact feature ma
tching is attempted and no initial approximate transformation is provi
ded. The method is naturally decomposed into two stages of initial est
imation and final refinement of the transformation. A hierarchical tri
angulation-based surface representation provides an efficient way to s
elect the control points at which the alignment of the two surfaces is
to be evaluated. This representation also permits the selection of a
manageable number of initial transformations among which at least one
is to be in the parametric neighborhood of the actual transformation.
Previous techniques are compared and their adaptation into an integrat
ed method makes it possible to go beyond the identified limitations. E
xperimental results show that the computed transformation between two
views of a complex multipart object may provide angles of rotation wit
hin a fraction of a degree of the actual ones. (C) 1995 Academic Press
, Inc.