MODELING AND NUMERICAL OPTIMIZATION FOR T HE RECONSTRUCTION OF A POLYHEDRON THROUGH ITS GENERALIZED GAUSSIAN IMAGE

The representation of an object by means of its extended Gaussian imag e, which we write E.G.I., occurs in many computer vision problems. A c onvex object is fully specified by its E.G.I. In this work, we present an effective algorithm to reconstruct the surface shape of a convex p olyhedron from its extended Gaussian image. This reconstruction is don e by solving a convex optimization problem. The general scheme of the algorithm is given along with some comments on numerical aspects.