Abstract The accuracy of phase shift extraction has a significant influence on measurement results in surface micro-topography interferometry. Phase shifting errors are mainly caused by nonlinearity of the employed phase shifter, environmental turbulence, camera imperfection and so on. In this paper, a general algorithm based on Lissajous figures and ellipse fitting is proposed for extracting the phase distribution from a set of phase-shifting interferograms with random noise. Two sets of pixels with π ∕ 2 phase difference in all the investigated interferograms are selected and used for ellipse fitting. Both numerical simulations and optical experiments have proven the validity, rapidity, and accuracy of the proposed method. Experiments show that the proposed method is a general phase extraction method, which can work for straight fringe patterns, circle fringes patterns and other anomalous features.