Detection of interest points on 3D data: Extending the Harris operator

Abstract

We consider a problem of interest point detection, i.e. location of points with standing out neighborhood, for a 3D mesh data. Our research is motivated by the need of general, robust characterization of a complexity of the mesh fragment, to be used for mesh segmentation and description methods. We analyze the reasoning behind traditional Harris operator for 2D images [4] and propose several possible extensions to 3D data. We investigate their performance on several sets of data obtained with laser digitizer.