We propose a high signal-to-noise extended depth-range three-dimensional (3D) profilometer projecting two linear-fringes with close phase-sensitivity. We use temporal phase-shifting algorithms (PSAs) to phase-demodulate the two close sensitivity phases. Then we calculate their phase-difference and their phase-sum. If the sensitivity between the two phases is close enough, their phase-difference is not-wrapped. The non-wrapped phase-difference as extended-range profilometry is well known and has been widely used. However as this paper shows, the closeness between the two demodulated phases makes their difference quite noisy. On the other hand, as we show, their phase-sum has a much higher phase-sensitivity and signal-to-noise ratio but it is highly wrapped. Spatial unwrapping of the phase-sum is precluded for separate or highly discontinuous objects. However it is possible to unwrap the phase-sum by using the phase-difference as first approximation and our previously published 2-step temporal phase-unwrapping. Therefore the proposed profilometry technique allows unwrapping the higher sensitivity phase-sum using the noisier phase-difference as stepping stone. Due to the non-linear nature of the extended 2-steps temporal-unwrapper, the harmonics and noise errors in the phase-difference do not propagate towards the unwrapping phase-sum. To the best of our knowledge this is the highest signal-to-noise ratio, extended depth-range, 3D digital profilometry technique reported to this date.
PSNet: a deep learning model based digital phase shifting algorithm from a single fringe image
National Engineering Laboratory for Integrated Aero-Space-Ground-Ocean Big Data Application Technology, College of Computer Science, Northwestern Polytechnical University（西北工业大学计算机学院）
