Efficient Surface Geometry Measurement by Means of Optical Fringe Projection

Optical measurement provides a means for automated in-process inspection of engineering surfaces. This paper presents an efficient technique to determine the three dimensional geometry of a curved surface by means of optical fringe projection that is different from the well known structured light. The geometry measurement accuracy thus obtained greatly relies on the accuracy of fringe center detection. To locate fringe centers more accurately, subpixel resolution is developed through an optimum reconstruction filter. To compare the measurement accuracy and repeatability between employing one-pixel and subpixel resolutions, a cylinder of a known diameter was used. Computationally efficient algorithms to determine the third dimension are presented (in addition to the two dimensions obtained from a camera). A turbine blade was used as a typical curved surface for measurement. To verify the developed algorithms, the optical measurement data were compared with those obtained from a coordinate measuring machine (CMM). The comparison indicates that they are generally in agreement except at some places where the surface curvatures drastically change.