To meet the demand for high-precision detection and location of surface defects in industrial inspection, this paper proposes a defect feature reconstruction method. A texture camera is added to the three-dimensional (3D) reconstruction system of binocular grating projection to realize the texture mapping of the reconstructed point cloud model. The 3D reconstruction of surface features can be realized using the feature regions extracted from the image captured by the texture camera. Since the object needs to be reconstructed from multiple perspectives, a precision turntable is introduced. The rotation axis calibration method is used to obtain the projection relationship between the texture image and the point cloud data at different rotation positions, and the judgment method based on the distance criterion is employed to eliminate the occluded part of the point cloud. Finally, the pixel coordinates of the texture image are located in three dimensions by four-quadrant proximity search and linear interpolation based on distance weighted average. The experiments complete the reconstruction of the pixel points in the marked defect contour in the image and realize the accurate size calculation and location of surface features. The reconstructed defect size and position are calculated which are then compared with the results measured by an optical image measuring instrument. The results show that the measurement error of the proposed solution for the 3D sizes and positions of defects is no more than 0.2 mm, and the areas of defects can be calculated more accurately.