Nersesian, GevorgSarvazyan, NarineKhachatryan, SurenCampana, StefanoFerdani, DanieleGraf, HolgerGuidi, GabrieleHegarty, ZackaryPescarin, SofiaRemondino, Fabio2025-09-052025-09-052025978-3-03868-277-6https://doi.org/10.2312/dh.20253359https://diglib.eg.org/handle/10.2312/dh20253359Armenian monuments are rich in carved stone inscriptions. These inscriptions serve as vital records of cultural and linguistic heritage, offering insights into the lives, beliefs, and traditions of Armenians during the Middle ages. However, detecting and comprehending these inscriptions pose significant challenges. Due to weathering, vandalism, erosion, and the complexity of ancient scripts, many of these texts remain unreadable. Yet, the few existing studies indicate that deciphering these messages from the past is feasible with technological advancements. In the present project we study a unique, newly created and unex- plored collection of digital twins of Armenian tapanakars (tombstones) and khachkars (cross-stones) focusing on hierarchical segmentation of the images using the detected geometrical and statistical features. The results are applied to character classi- fication and the accuracy of the generated images is estimated. Since the detection stage of the algorithm is universal for any kind of shapes, it opens up new research avenues that extend beyond text recognition alone. The same pipeline can be adapted to identify decorative motifs, geometric symbols, and other visual patterns commonly found on tapanakar surfaces.Attribution 4.0 International LicenseCCS Concepts Computing methodologies → Computer vision problems; Supervised learning by classification; Information systems → Optical character recognition; Document representationCCS Concepts Computing methodologies → Computer vision problemsSupervised learning by classificationInformation systems → Optical character recognitionDocument representation10.2312/dh.202533595 pages