Deep absolute phase recovery from single- frequency phase map for handheld 3D measurement

Fringe projection profilometry is an effective technique for handheld three-dimensional shape measurement because of its high resolution and accuracy. However, it is hard to precisely achieve rapid measurement for complex scenarios using traditional fringe analysis methods without additional patterns or camera(s). Recovering the absolute phase from single-frequency fringe patterns of complex scenarios is challenging and still not well-addressed. This paper proposed a learning-based approach to address the unwrapping of one single-frequency phase map. A fully convolutional neural network is introduced to learn the mapping from wrapped phase to fringe-order without additional images under the constraint of the boundary fringe-order. This network is designed as a lightweight one and can operate in real-time for high-resolution phase maps. Experiments on a real large-scale dataset demonstrated that the presented method could unwrap single-frequency phase maps of handheld 3D measurement with motion blur, phase discontinuity, and isolated regions.