An efficient computational phase extraction from arbitrary phase-shifted fringes patterns

摘要

Phase-shifting is an important technique for phase retrieval in interferometry and three dimensional profiling by fringe projection, which requires a series of intensity measurement with known phase-steps. Usual algorithms are based on the assumption that the phase-steps are evenly spaced. In practice, the phase steps are not evenly spaced or exactly determined or measurement, which leads to errors in the recovered phase. Based in this fact, some iterative algorithms have been proposed, e.g. Advanced Iterative Algorithm, which is a self-calibration algorithm for phase retrieval, however, it converges slowly. In this work, we propose an efficient-computational strategy for implementation of the AIA algorithm. The proposal consists of two steps: a method to reduce the number of iterations, and the use of high-performance computing techniques to reduce the computation time at each iteration. The strategy is validated using synthetic and real data. Results show a drastic reduction in the number of iterations and increased performance.