A similarity measure for non-uniformly sampled multi-resolution terrain data using the open-ball operator

This paper presents a new method of measuring the correlation (similarity) between non-uniformly sampled digital elevation models. The similarity measure has been calculated for different levels of detail using the multi-resolution property of wavelets. To obtain a feature representation from a terrain data, we extract a set of feature matrices using the open-ball-operator (OBO). The OBO is capable of not only mapping the information on convexities and concavities to a feature matrix but also providing the invariance with respect to shift, scale, and rotation to the terrain data base.