Photogrammetric material mapping of heritage facades under constrained survey conditions

Abstract This study investigates facade-scale photogrammetric material mapping of architectural heritage under constrained survey conditions. High-resolution orthophotos derived from terrestrial image acquisition were analyzed using a color-encoded Pixel Texture Segmentation (PTS) framework comprising eight material classes. To complement qualitative inspection, a region-of-interest (ROI)-based quantitative validation scheme was implemented using confusion matrices and intersection-over-union (IoU) metrics. Four facades of the Adana Ulu Mosque were examined to evaluate facade-dependent variability in segmentation performance. The results indicate pronounced orientation-related differences in classification stability. Dominant masonry classes, particularly clean and weathered limestone, achieved moderate to high IoU values across multiple facades. In contrast, visually similar or spatially limited materials, such as brick infill, decorative bands, and metal elements, exhibited systematic misclassification patterns. These findings define the operational reliability limits of texture-based segmentation and highlight the importance of transparent facade-scale validation in digital heritage documentation under constrained survey conditions.