To reduce the effect of large number of invalid data in fringe patterns on calculation results in the measurement of high dynamic flows using interferometry or Moiré deflectometry, an effective phase extraction method was proposed based on the wavelet transform using complex Morlet wavelet. Through choosing proper parameters for the wavelet, the maximum modulus of the wavelet transform of a pattern was proportional to its modulation factor, then it was used as a weighting factor of a phase unwrapping algorithm based on weighted least squares to guarantee reliable phase unwrapping. In the process of wavelet ridge routing, a simple algorithm based on curve fitting was proposed to substantially reduce the routing time and the precision deterioration due to image noise. In simulation experiment, the relative error between calculated phase and actual phase is less than 0.01%. In internal wave measurement experiment, the method is used to analyze the fringe patterns, and the density gradient measurement precision has reached 5×10-6 g/cm4. It shows that the method can reduce the effect of invalid data on analysis precision effectively and can obtain reliable unwrapped phases.