Accurate calibration for fringe projection profilometry based on an improved subpixel mapping with local gray distribution.

Fringe projection profilometry is an efficient and accurate technique for three-dimensional (3D) measurement to calibrate a camera and projector setup. The feature centers of circles on a calibration board are extracted on the camera image plane and mapped to the projector image plane during the calibration procedure. The accuracy of the mapping between camera pixels and projector pixels is crucial to the calibration accuracy, which directly affects the measurement precision of the system. In this paper, we investigate an improved subpixel mapping with local gray distribution from the camera to the projector. The mapped pixels and their gray values are regarded as a set of 3D grayscale space points. The subpixel coordinates of the feature centers on the projector image plane are obtained by directly processing the 3D points. The entire procedure retains the subpixel precision. Calibration experiments were designed to verify the feasibility of our calibration method, which was compared to three existing methods. The reprojection errors and object-space errors were used to evaluate the calibration accuracy of the methods. Additionally, measurement experiments of displacement and in-plane distance were employed to verify the calibration results of the methods. Compared to the three existing methods, we believe our method can improve the calibration accuracy for fringe projection profilometry.