Undeformed superquadrics are volumetric modeling primitives with an extensive shape vocabulary that are described by only 5 parameters. Fitting these models viewpoint invariantly to range data enables classification based on the superquadric parameters. However, current recovery routines show several limitations, especially when the algorithms are applied to range images instead of true 3D images. In this paper problems with the common superquadric recovery procedure are identified and solutions are presented.

A statistical model, describing noise-disturbed invariants extracted from a surface patch of a range image, has been developed and applied to region based pose estimation and classification of 3D quadrics. The Mahalanobis distance, which yields the same results as a Baysian classifier, is used for the classification of the surface patches. The results, compared with the Euclidean distance, appear to be much more reliable.