Research Of Windowed Fourier Transform Based Time Series Phase Extraction Algorithm In Crystallization Process

Windowed Fourier Transform based Time Series phase extract algorithm is an useful method with high accuracy in areas of phase extraction from fringe pattern. But the size of window and other parameters are important to the precision of the phase extraction. The selection of size of window in the traditional phase extraction process depends on the experience and there is no systematic error analysis on the phase extraction precision of each parameter.The following aspects are presented in this paper:(1).We do a series of simulation based on the parameters in the Windowed Fourier Transform based Time Series phase extract algorithm, and investigate the affection of window size selection by ratio of signal to noise, frequency interval and time interval, the signal shape parameters from the error analysis area in the phase extraction process. The result showed that the ratio of signal to noise is crucial to the precision of phase extraction and the selection of the best window size. Frequency interval and time interval make no sense to the precision of phase extraction and the selection of the best window size to some extent. The signal shape parameters affect the precision of phase extraction and the selection of the best window size through changing the ratio of straight line to curve. (2).We conclude the area of best window size of each parameter and get an best window size selection area which is applied to the crystallization process on the basis of our former study, and we make a validation, the sodium chloride natural evaporation crystallization experiment, to certify the precision of the phase extraction. The result is positive.