Inexact Correspondences

Two or more 3D data sets are unlikely to be acquired such that the 3D data points exactly correspond. For example, the sampling density might be different due to scanner settings, scanner distance or surface slopes. Thus, a point or feature in one data set will not have an exact correspondence in another data set, yet some sort of alignment between the elements in the two data sets will be needed. Hence, methods for finding appropriate correspondences is being researched (13,49,7,19). However, it is observed that registration errors are a function of interpoint distances and thus, the resolution of the scanning. The smaller such distances, the larger the potential for accurate registration. This suggests that registration accuracy may be improved from fitted surfaces.