The paper presents a method to identify defects from fringe patterns. In the proposed method, the phase derivatives are computed from a fringe pattern using the two-dimensional Pseudo-Wigner–Ville distribution. Since the phase derivative varies rapidly in the vicinity of the defect, the relative change in the derivatives for the normal and defect-containing fringe patterns is compared with respect to a preset threshold to identify the defect in the fringe pattern. The robustness of the method for detecting defects of various sizes and at different noise levels is shown using simulated fringe patterns.