Accurate fringe projection profilometry using instable projection light source

Abstract Fringe projection profilometry (FPP) usually uses a projector with an instable light source due to the unsatisfactory driving power. As a result of the fluctuations of the projection light source, the fringe captured by a camera deviates from its ideal distribution, from which, the calculated phase has a random phase error, thus reducing the 3-D measurement accuracy. We found that the random phase error influences both the system calibration and the 3-D reconstruction, where the influence on the system calibration has not been previously noticed. In this paper, we formulate the phase error model of the phase-shifting algorithm caused by the fluctuations of the projection light source, and analyze its interference on calibration and reconstruction, respectively. From which, an accurate FPP using a combination of accurate calibration and accurate 3-D reconstruction is proposed, which are performed by using a deep learning-based phase calculation method. The provided experimental results verify that the proposed FPP can obviously improve the measurement accuracy.