Improving the performance of fringe projection profilometry by accurate geometric modelling and system calibration

This thesis presents a number of new techniques to improve the accuracy of threedimensional (3D) shape measurement using fringe projection profilometry (FPP). Firstly, approaches for camera calibration are systematically investigated, and a new method is proposed to achieve sub-pixel mapping between the liquid crystal display (LCD) and the charge-coupled device (CCD). This approach uses vertical and horizontal fringe patterns and centrelines displayed by a LCD as the calibration board. A set of pixels on the designed circle edge are then mapped from the LCD onto the CCD. The circle centres are extracted on the CCD by the least-squares circle fitting technique. In contrast to existing approaches, which only use the horizontal and vertical fringe patterns, the new method is more flexible and accurate.