Measuring the three-dimensional (3D) shape of metal parts is essential for evaluating their processing quality. The 3D reconstruction scheme based on structured light is affected by the bright reflections from the metal parts, which lead to reduced signal-to-noise ratio and loss of data. To solve the highlighting problem in the detection of metal parts, an adaptive fringe projection measurement scheme is proposed herein and is applied to the measurement of shape and dimension of metal surface. The highlight area of the image is detected by a combined thresholding method and color-space transformation method. The transformation relationship between the camera and projector is obtained by a projection-camera dual-view search method based on an epipolar constraint. The modulation parameters are generated by a scheme based on the bidirectional reflectance distribution function (BRDF) optical imaging model. The BRDF-based scheme realizes pixel-level modulation of the projection fringe. In addition to reducing the number of projections and improving the measurement efficiency, the compensation rate of invalid area obtained by the proposed method is 51.3% in the highlight area measurement results, and the measurement accuracy is improved up 77.4%.