Robustness of novel surface invariance to geometric transformation

In this paper we explore the novel geometric invariance on surfaces based on the set of invariant normal vectors that are relatively preserved under geometric transformations, are local, intrinsic and computed from the differential geometry of the surface. To reduce the sensitivity of the computation of the geometric invariance to noise, we use a B-Spline surface representation that smoothes out the surface prior to the computation of these invariant points. The robustness of the geometric invariance is shown for a variety of geometric transformation. The result is very promising.