Recognising decorations in archaeological finds through the analysis of characteristic curves on 3D models

In the analysis of archaeological finds, it is important for archaeologists to identify their style, origin, period, etc. to allow their correct classification. In the digital era, the development of automatic techniques to measure the peculiar characteristics of archaeological finds would be of great help in this activity. Considering that ancient artefacts are very often incomplete, consumed, degraded, if not consisting of simple fragments, geometric details, such as decorations, visual motifs, patterns, are more useful for their analysis than global characteristics. These patterns are usually composed by characteristic curves arranged in a regular way, as in a Greek fret or a floral band. Here we propose the recognition of characteristic curves on 3D models of archaeological artefacts, identified by a set of characteristic points. We approximate these curves with known curves to provide localisation and quantitative measurement of the characteristic features used as decorations or patterns of the digital models of ancient objects. To solve this problem, we adopt a generalised version of the Hough Transform (HT). In addition, we introduce new rules of composition and automatic aggregation of the characteristic curves, not limiting the recognition to a single curve at a time and supporting an automatic annotation of the fragment digital model. (c) 2020 Elsevier B.V. All rights reserved.