Rational Gaussian surface fitting
Image Registration and Fusion Systems      

If points representing pixels in an image along with their intensities are given, or if points representing height values in a scene along
with their UTM coordinates are available, a parametric surface can be fitted to the points to recover the underlying geometry. If the
points represent volumetric data and the points can be parametrized (e.g., cylindrically or spherically), again, a parametric surface
can be fitted to the points.

This software implements the rational Gaussian surface fitting method. Rational Gaussian surfaces are parametric surfaces. If data
points are in 2-D, the surface parameters can be set proportional to the x and y coordinates of the points. Otherwise, there is a need
to determine and provide the parameters at the points. Open, half-closed (cylindrical) and closed (spherical) parameters are
accepted.

In this surface fitting method the standard deviation of the Gaussians can be used as the smoothness parameter. As the standard
deviation of the Gaussians is increased more noise or local details are smoothed, creating a smoother surface. The software allows
the user to interactively change the smoothness parameter and view the constructed surface.

An example of surface fitting by this software is given below.
Fig. 1. Top row shows the original points as well as the points in triangulated and shaded forms. Middle row shows surfaces at
different smoothless levels approximating the points. Third row show overlaying of the surfaces with the original triangulated points.
To obtain a license for this surface-fitting software,
                                                            please follow the link =>