A study on the discrimination and identification of ancient glassware types based on cluster analysis

Abstract

The classification and identification of ancient glass artifacts is complicated due to the exchange of internal elements, which can be influenced by a multiplicity of different factors, such as weathering, erosion, and so on. To classify and identify the types of glass artifacts, this paper uses cluster analysis (K-Means) to study the main chemical composition of high potassium glass and lead-barium glass. It is found that the chemical composition of different glass ( potassium oxide, barium oxide, lead oxide) and subclasses types are significantly different, and the content of lead oxide in lead-barium glass also has a large difference. Using the clustering centers obtained by cluster analysis (K-Means), the eight unknown types of glass were first determined visually and briefly based on the different proportions of potassium oxide, barium oxide, and lead oxide. As for accurate classification, the Euclidean distances between the unknown components and the sample centroids were calculated separately, and the shortest Euclidean distances were identified by comparison, and then categorized for identification.