Automated classification of archaeological ceramic materials by means of texture measures

TY - JOUR

T1 - Automated classification of archaeological ceramic materials by means of texture measures

AU - Hein, Irmgard

AU - Rojas-Domínguez, Alfonso

AU - Ornelas, Manuel

AU - D'Ercole, Giulia

AU - Peloschek, Lisa

PY - 2018/10

Y1 - 2018/10

N2 - We explore the use of image analysis techniques for the classification of archaeological ceramic materials according to one aspect of their petrographic characterization. Specifically, we study the use of Gabor filter-based texture features, Laws' texture measures, and Haralick's texture measures for automated classification of a set of archaeological ceramic samples from two different Egyptian source materials: Marl clay and Nile clay. The motivation behind this work is the desire to pioneer the introduction of fully automated methods for pattern classification into the domain of archaeological science, where these can be extremely useful. The texture features are all extracted in a completely automated fashion and the classification is performed via a simple classification algorithm, the k-NN classifier. An accuracy of nearly 74% was obtained with base on the Laws' texture features.

AB - We explore the use of image analysis techniques for the classification of archaeological ceramic materials according to one aspect of their petrographic characterization. Specifically, we study the use of Gabor filter-based texture features, Laws' texture measures, and Haralick's texture measures for automated classification of a set of archaeological ceramic samples from two different Egyptian source materials: Marl clay and Nile clay. The motivation behind this work is the desire to pioneer the introduction of fully automated methods for pattern classification into the domain of archaeological science, where these can be extremely useful. The texture features are all extracted in a completely automated fashion and the classification is performed via a simple classification algorithm, the k-NN classifier. An accuracy of nearly 74% was obtained with base on the Laws' texture features.

KW - Automated image processing

KW - Ceramic analysis

KW - Gabor filters

KW - Image analysis

KW - Pattern classification

KW - Texture extraction

U2 - 10.1016/j.jasrep.2017.12.032

DO - 10.1016/j.jasrep.2017.12.032

M3 - Journal article

AN - SCOPUS:85039157769

SN - 2352-409X

JO - Journal of Archaeological Science: Reports

JF - Journal of Archaeological Science: Reports

ER -