Search

Constant radius geometric features are a common type of manufacturing features of ancient ceramics. They are obtained by a sweeping action of a tool, which leaves negative or positive traces characterized by a cross-section with one or more constant radii. The automatic recognition and dimensional characterization of these features could be useful for understanding the technology used to manufacture ceramics. Thus, a new perspective in archaeological investigations can be furnished. For this purpose, in this paper a new computer-based methodology suited to segmenting constant radius geometric features and measuring their dimensional parameters is proposed. Starting from a 3D discrete geometric model of the ceramic, the region pertaining to these features is determined and measurements of their radii are performed. Due to the uncertainties of various sources, which affect the investigated object, the required process is not trivial. In order to solve this problem, the segmentation phase is conducted using a non-conventional logic suitable for exploring the object with a fuzzy sensitivity, and the measurement is performed by a robust fitting method applied to the segmented entities. The methodology has been tested in the identification of embossed decorations of an ancient olla. The combined effects of the feature segmentation process together with the measure detection approach on the obtained results are critically analyzed and discussed.