Data fusion for cylindrical form measurements

Abstract

For high-precision form measurements of cylindrical workpieces form profiles such as roundness and straightness profiles are independently acquired via a bird-cage strategy. The 3D point cloud reconstructed by fusing these intersected profiles is meaningful in dimension and form assessment for cylinder, since enhanced accuracy can be achieved by fusion results. Moreover, it plays an important role as the input to other calculations. However, these data cannot be accurately aligned in form reconstruction, due to random absolute offsets in profiles and a lack of absolute positions. Therefore, we propose an approach to data fusion of these profiles to reconstruct cylindrical form. The uncertainties of the fused profile are evaluated, taking an individual contribution of a single profile and a global contribution of all profiles into account. The associated uncertainties are propagated using the Monte Carlo method. Experimental study results indicate that the data fusion procedure improves the accuracy of available datasets. After fusion, all available data points are capable of being used in the form assessment.