Window size selection in windowed Fourier transform for phase retrieval

Windowed Fourier transform (WFT) has been proved to be one of the most effective algorithms for extracting phase from various optical fringe patterns. The accuracy of phase retrieval using WFT depends on the preset window size, but how to determine an appropriate window size is still a problem. In this paper, a mathematical relationship between the optimal window size and the phase derivative is deduced and several advices on window size selection in practical applications are proposed. The correctness of the proposed window size selection criterions is demonstrated using both numeral simulations and real experiments. Comparison of phase values retrieved by conventional approaches and WFT shows that the WFT with an optimal window size can give more precise results.