Discussions on phase-reconstructing algorithms for 3D digitizing structure-light profilometry

For the 3D digitizing structure-light profilometry, five kinds of phase reconstructing (PR) algorithms, including Schofield's S-FFT (with 4 pairs of FFT & Inverse FFT operators) algorithm, Volkov's deterministic V-FFT PR algorithm (with only one pair of FFT operators and one Inverse FFT operator), quality-guided (QG) method, weighted least-square (WLS) method and a proposed Discrete-Cosine-Transform-Based (DCT-B) algorithm have been implemented and discussed in detail. The computer-simulated data show that the WLS and V-FFT enjoy the least phase error, but the QG algorithm achieves the trade-off between reliability and computational time. The S-FFT together with V-FFT method almost fails to deal with the experimental data with abrupt phase variance. Among the other three algorithms, the DCT-B algorithm shows its advantage over QG and WLS algorithms in settling these problems. Considering phase error and computational time, it could win the priority in PR. The experimental results also demonstrate that the proposed DCT-B algorithm could be firstly considered in the application of structure-light profilometry, especially to process the abrupt-phase maps.