Abstract: Structured light fringe projection methods have been commonly used for three-dimensional (3D) shape measurements. However, for objects with large surface reflectivity range, saturation of the captured fringe pattern images leads to the phase information unavailable. The traditional high dynamic range scanning (HDRS) technique is complicated and time-consuming. An adaptive fringe-pattern projection method is proposed. Firstly, a sequence of phase-shifted fringe patterns with high gray level is projected onto surface of the measured object, which is used to predict and mark the overexposed points. Subsequently, the maximum input gray level (MIGL) of each saturated pixels is determined by the nonlinear least squares (NLS) fitting method based on the captured images with middle exposure level. Finally, the adapted fringe patterns are projected for phase calculation and 3D shape recovery. The experimental results demonstrate that the proposed method effectively achieved 3D shape measurement of object by avoiding image saturation in high-reflective surface regions. The simulation error of the proposed method is within the range of 0.02 mm. Compared with this, the actual measurement error is 0.14 mm, and the compensation rate for the over-exposure points can reach 99% in the actual experiment.