An Automatic Three-Dimensional Fuzzy Edge Detector

Abstract

Three-dimensional object analysis is of particular interest in many research fields. In this context, the most common data representation is boundary mesh, namely, 2D surface embedded in 3D space. We will investigate the problem of 3D edge extraction, that is, salient surface regions characterized by high flexure. Our automatic edge detection method assigns a value, proportional to the local bending of the surface, to the elements of the mesh. Moreover, a proper scanning window, centered on each element, is used to discriminate between smooth zones of the surface and its edges. The algorithm does not require input parameters and returns a set of elements that represent the salient features on the model surface. This method is general enough, returns representative structures of the object, as edges, and can be considered as a pre-processing step for further applications, such as 3D compact representation, matching and recognition