Image processing and hypermedia tools for archaeological classification

The primary aim of this thesis was to produce an improved methodology for classifying archaeological artefacts on the basis of shape information.The need for artefact classification was identified early in the history of archaeology, but until the 1950s it was carried out solely by hand. Several numerical methods for artefact classification have been devised since that date, but none have passed into general use. I start by examining the problem itself and previous attempts to produce shape-classification methodologies to solve it. It is shown that the difficulties encountered in the past lay in applying the methodology to the problem. There are three main areas of difficulty: defining the problem of shape based classification itself; devising a suitable shape measure; and deciding on the most appropriate way of analysing these measurements. These three difficulties are tackled in the first part of the thesis. The remainder of the thesis, consisting of two case studies, describes the application of the methodology to real-life archaeological problems.The first case study, an analysis of Cretan pithoi, was undertaken primarily as a test of the methodology, to establish its precision and accuracy. The second case study, an analysis of Attic black-figure pottery, demonstrated the methodology's ability to work with any kind of shape. In addition to analysis of the artefact's profiles, the painted surface decoration of the vases was also analysed, to explore the differences between decoration painted by different individuals.