Acquiring decorative patterns from glossy ceramic surfaces is challenging because specular reflections often obscure fine details and reduce the reliability of subsequent digital analysis. Although existing highlight-removal methods, including data-driven and single-image enhancement approaches, have improved restoration quality in generic scenes, they are not fully suited to glossy ceramic documentation because they often rely on scene priors, large paired datasets, or post hoc enhancement alone while paying limited attention to acquisition-side optimization for reflective cultural objects. This article presents a portable multimodal imaging system and a processing framework for ceramic pattern acquisition, highlight suppression, and acquisition optimization. Multimodal images captured under different illumination and polarization configurations are first geometrically registered, after which specular regions are localized by jointly exploiting polarization and intensity cues, followed by highlight suppression and perceptual appearance restoration to improve pattern visibility while preserving visual authenticity. Experimental results indicates that warm illumination with 0° polarization is more suitable for warm-toned ceramics or ceramics with large-area patterns, whereas uniform illumination with 45° polarization is more suitable for cool-toned ceramics and ceramics with sparse patterns; additionally, cool illumination with 90° polarization yields the highest average score across the dataset, indicating stronger robustness across diverse samples. The proposed system is portable, supports wireless image transmission, and integrates adjustable illumination with a servo-driven polarizer, thereby providing a practical solution for high-quality digital documentation of glossy ceramic patterns.
Lei et al. (Thu,) studied this question.