We introduced a new parametric warp we called NURBS-Warp. It was derived by analyzing the classical BS-Warp, that we showed is based on the two-way tensor product of bivariate B-Splines. As a first contribution, we showed that the BS-Warp is intrinsically affine: it does not model the effect of perspective projection. Our NURBS-Warp, based on the tensor product of NURBS, is an extension of the BS-Warp. It models perspective projection, and thus copes with more complex deformations with less control points, as our experimental results show. An estimation procedure from point correspondences was given. It allowed us to demonstrate the representational power of our warp compared to the BS-Warp, and against various factors such as noise, surface shape and number of control points. However, it might be interesting for future work to study how classical robust and pixel-based methods can be applied to our NURBS-Warp.

Contributions to Parametric Image Registration and 3D Surface Reconstruction (Ph.D. dissertation, November 2010) - Florent Brunet
Webpage generated on July 2011
PDF version (11 Mo)