摘要：This paper pioneers a new computational approach for the study of changes in shape of objects across time. Previously, such a study was undertaken by scholars using a purely visual approach and relied on images of objects or in-person observations. This paper’s approach is based on 3D scans of historical artefacts. Sample points are extracted from these 3D scans and the distance between analogous points across different objects is computed using an approximation of the Wasserstein metric, namely the Sinkhorn distance. In this paper, the approach is demonstrated on a small set of ancient Greek vessels of the Krater, Pelike, and Kylix types, as the variation in their shapes across time is well known to archaeologists. Results offer, for the first time, a way of quantifying differences between objects. Benefits of this approach lie in its ability to quantify change, to study complex 3D material, and to analyse large datasets of objects, opening the possibility of constructing new large-scale studies of object shape across time and geographical regions. These have a range of applications in art history, archaeology, digital humanities, museology and extended reality studies.
