3749 - Analyzing Archaeological Masks through Quantitative Methods and Computer Vision

This paper presents a project focused on developing new methodologies for the analysis of archaeological masks. The project applies quantitative techniques such as Spectral Clustering to identify significant classes in a collection of masks. We demonstrate that Spectral Clustering yields better results in isolating meaningful groups than more traditional methods such as k-means and numerical taxonomy through single and total linkage techniques. A good clustering is a starting point in identifying the key attributes that can define a class and could later lead to a better understanding of a "style". This paper also applies techniques from Computer Vision and three-dimensional shape matching to rank similarities between great numbers of artifacts in an automatic way. The work includes 3D scanning techniques to produce three dimensional models of artifacts, as well as the production of a recognition system to analyze 3D shapes. Our interest is to introduce the archaeological community, especially those colleagues studying masks, to a range of computer tools that can be applied to a better understanding of these collections. We use the collection of 162 masks recovered from the Sacred Temple of Tenochtitlan as a study case. This includes some Teotihuacan style items as well as other masks from the Mezcala region. The methodologies proposed, however, are intended to be generic and can be applied to any collection of masks.

diego.jimenez61@gmail.com, src@cimat.mx, omendoz@cimat.mx 

Keywords: Stone Masks, 3D Shape Analysis, Archaeological masks, Computer Stylistic Analysis

Author: Jimenez-Badillo, Diego (Instituto Nacional de Antropologia e Historia, Mexico, Mexico / Mexiko)
Co-Author: Ruíz-Correa, Salvador; Mendoza-Montoya, Omar (Centro de Investigaciones en Matemáticas, Guanajuato, Mexico / Mexiko)


University of Vienna | Dr.-Karl-Lueger-Ring 1 | 1010 Vienna | T +43 1 4277 17575