Two approaches, single decomposition approach and multi-decomposition approach,using discrete wavelet transform for extracting phase in phase profilometry are proposed. In the former approach, wavelet multi-resolution analysis is chosen to perform discrete wavelet transform of the pattern after converting original signal into the analytic form. The phase distribution is recovered from every coefficient whose modulus reaches maximum in the scale direction at the position. In the latter one, an interpolation to original signal is used to vary the sampling rate and multi-resolution analysis is reused to acquire redundant detail coefficients. The first approach is suitable for sinusoidal fringe pattern. In contrast, the second one is more robust to noise and capable of handling quasi-sinusoidal fringe pattern. Compared with the method employing continuous wavelet transform, these two approaches have 90% and 50% increases in computing speed respectively for a 512 pixel×512 pixel image, and are more appropriate to discrete signal processing. The numerical simulation and experiment have proved the validity of the approaches.