| Abstract: | We present a new technique for the recovery of local phase from multiple phase-stepping fringe images that uses adaptive quadrature filters constructed by use of Bayesian estimation theory and complex-valued Markov random fields as prior models. It is shown that with this technique it is possible to perform accurate phase reconstructions even for extremely noisy fringe images and that the performance of this technique is nearly independent of the particular noise model, as long as the noise spectrum is wideband. |
