Adaptive pattern fusion for multi-reflectivity objects in fringe projection profilometry

Measuring the shape of objects with varying reflectivity is a challenging task due to the poor imaging quality. In this paper, we introduce a novel phase error model and a pattern fusion framework. First, a structured light illumination model with various light components as well as a phase error model based on the analysis of high-order harmonics of the captured patterns is proposed. After that, based on these models, we propose a novel quality criteria that leverages the modulation intensity and ratio of the direct light component. This criteria is used for pattern fusion and can minimize phase error. Extensive experiments demonstrate that our proposed scheme effectively preserve the reliable component in pattern fusion, and yields higher accuracy and less distortions in the depth maps.