Abstract

For i.i.d. data $(x_i, y_i)$, in which both $x$ and $y$ lie on a sphere, 
we  consider flexible (non-rigid) regression models, in which 
solutions can be obtained for each location of the manifold, 
with (local) weights which are function of distance. By considering 
terms in a series expansion, a ``local linear'' model is proposed for 
rotations, and we explore an iterative procedure with connections to 
boosting. Further extensions to general shape matching are discussed.