Work

Understanding the Style Differences of Ancient Ridergods Rock Reliefs with Feature Extraction: A Preliminary Research

Year: 2026

Type: article

Abstract: This study applies a computational approach to clustering grayscale images of rock reliefs from the Caria, Pisidia and Kibyratis region to detect the number of potential master carvers and/or schools responsible for these ancient artifacts. The model uses a dataset consisting of only 100 images and employs a feature extraction approach combining edge detection, texture analysis, shape analysis, and feature descriptors. We evaluate clustering performance using multiple metrics including silhouette score, Davies-Bouldin index, and Calinski-Harabasz index. The optimal clustering solution identified probably two distinct stylistic groups among the reliefs, corresponding to the presence of two different master carvers, schools or workshops as previously considered by human experts. This computational approach complements traditional archaeological methods and provides a more objective and analytical basis for stylistic classification of ancient rock reliefs as a preliminary proof of concept.

Source: Open Archaeology

DOI: 10.1515/opar-2025-0076

Authors: Eray Dökü, Deniz Kayıkcı, Nurhan Kılıçkaya

Institutions: Burdur Mehmet Akif Ersoy Üniversitesi, Universitat Autònoma de Barcelona, Pamukkale University

Language: English

Topic: Image Processing and 3D Reconstruction

Subfield: Computer Vision and Pattern Recognition

Field: Computer Science

Domain: Physical Sciences

Open Access status: gold

Estimated APC paid: $848

Funder: Universiteit Leiden