Dual Component Linear Interpolation of View Spheres.
N. A. Thacker · 2009
This paper describes an approach for the representation of projected 3D edge features for purposes of view-based recognition and localisation of objects. It is based upon the representation of arbitrary configurations of features using geometric co-occurrence as Pairwise Geometric Histograms (PGH). Sets of these histograms can be shown to provide a complete representation for arbitrary configurations of fixed 2D projections of 3D shape. Topological considerations allow us to determine the local variation of these descriptors as a function of viewpoint and scale. We describe a mathematical model for the interpolation of correlated changes in these histograms, comprising two independent linear models for use during simultaneous view and scale matching. Mismatch and match distributions are provided to give a context for the accuracy of approximation. An assessment is made of the utility of this approach for object localisation (more results are presented in a companion paper).