To solve the issue of phase recovery from low-quality fringe images (where the low quality is caused by noise, non-linear intensity and surface reflectivity changes), an iterative phase-correction method is proposed. The high-phase-precision solution is achieved by regenerating the phase-shift-fringe images from the wrapped phase and performing iterative filtering. In general, compared with conventional methods, the proposed iterative phase-correction method has three main advantages: (a) it suppresses the influence of noise without interference from surface reflectivity; (b) it effectively improves the wrapped phase accuracy, thereby improving 3D measurement accuracy, and (c) it effectively recovers the absolute phase (determined by a multi-frequency heterodyne method) information of surfaces with very low or very high reflectivities, without the additional projection of phase-shift-fringe images. The effectiveness of the proposed iterative phase-correction method is verified by simulation and experiment. The proposed method, applied to binocularly structured light, can improve measurement accuracy and greatly improve the effectiveness of measurement of objects with drastic changes in reflectivity. Based on the calibration accuracy (without gamma calibration of the projector) described in this paper, the root-mean-square deviation of the measurement results for binocularly structured light can reach 0.0094 mm.