Wavefront phase retrieval is one of the most critical problems in adaptive optics. Here, phase retrieval by solving the transport of intensity equation using membrane vibration modes is proposed. Our study shows that the wavefront curvature sensing signal on the pupil can be expanded as a set of corresponding membrane vibration modes. The analytic expressions of the reconstructed phase are given. The coefficients of the functions are obtained by the integral over the pupil and boundary. Several representative Zernike circular and annular polynomials are respectively fitted by eigenfunctions and membrane modes in the absence of noise. In addition, wavefront recovery from noisy curvature data of the simulated atmospheric turbulence phase based on Zernike modes and Kolmogorov spectrum is demonstrated to verify the accuracy and robustness of the proposed method.