Automatic Classification of Archaeological Potsherds

During archaeological excavations, one of the most time consuming stages is the treatment of the great number of pottery sherds found on the site (labelling, drawing, measuring and classifying as related to the known object models). This step is also the most difficult one because it requires an extended knowledge of the characteristics of the already identified object models that can be found on a given archaeological site. As digitalization techniques have become financially reachable, new computerized solutions to this problem can help the archaeologists. We present in this paper a matching method between a sherd and a shape model, based on one hand on the use of Implicit Surfaces to obtain a distance metric and on the other hand on Genetic Algorithms in order to find the best possible position, relatively to the previous distance measure.