Inpainting for Fringe Projection Profilometry Based on Geometrically Guided Iterative Regularization

摘要: Conventional fringe projection profilometry methods often have difficulty in reconstructing the 3D model of objects when images so-called highlight regions due to strong illumination from nearby light sources. Within a region, pattern is overwhelmed by reflected light. Thus, information object, which originally embedded pattern, can no longer be retrieved. In this paper, novel inpainting algorithm proposed restore presence highlights. The method first detects based on Gaussian mixture model. Then, geometric sketch missing fringes made and used as initial guess an iterative regularization procedure for regenerating fringes. simulation experimental results show that accurately reconstruct even their large regions. It significantly outperforms traditional approaches both quantitative qualitative evaluations.