Application of Photometric Stereo 3D Reconstruction Algorithm in Non-destructive Detection of Artworks

Abstract

This paper examines the theoretical basis and potential applications of photometric stereo 3D reconstruction for surface documentation and deterioration assessment in artwork conservation. A methodological framework is developed by linking image acquisition under multiple illumination directions, robust weighted normal estimation, Poisson integration, and geometric feature analysis. Ancient Chinese paintings and calligraphy, lacquerware, and stone and jade artworks serve as representative categories for examining how reflectance, texture, and local relief affect reconstruction and interpretation. Conceptual figures illustrate the acquisition workflow and the representation of surface geometry. The theoretical analysis identifies reflectance as a factor affecting the reliability of normal estimation and explains how gradients, curvature, and height residuals can provide complementary cues for observing cracks and local deformation. The framework provides a methodological basis for non-contact documentation and analysis before conservation treatment. Its accuracy and range of application require further experimental evaluation for specific materials.