Minimizing LIBS damage in the analysis of decorative tiles using RGB data clustering

• Dual-domain clustering integrates RGB segmentation with LIBS spectral mapping. • Achieves high match between visual and chemical similarity across sample regions. • Cuts LIBS shots and data size by more than one order of magnitude. • Minimizes surface damage while preserving chemical representativeness. • Scalable workflow applicable to other layered cultural-heritage materials. Spectral analysis of cultural heritage materials offers valuable insights into the restoration and preservation of historical artifacts, revealing details about the materials used and the manufacturing techniques employed. However, given their historical and artistic significance, the extraction of elemental information from these fragile samples poses a unique challenge, as these objects must be examined using minimally invasive methods to prevent irreversible damage. Laser-induced Breakdown Spectroscopy (LIBS) is one such technique, providing a rapid and detailed elemental characterization. Yet, extensive LIBS analysis can still compromise the integrity of these delicate objects. In this work, a novel approach that integrates spectral and RGB data clustering to significantly reduce the number of LIBS measurements required is introduced. By segmenting the material into visually and chemically distinct clusters, this method enables targeted LIBS analysis using only a few representative shots per cluster, thus preserving the integrity of cultural heritage artifacts while still delivering reliable compositional insights.