A weighted least square based data fusion method for precision measurement of freeform surfaces

The trend towards product miniaturisation and multi-functionality constitutes a driving force for the application of complex surfaces in many fields such as advanced optics. The precision measurement of these surfaces should be carried out at multiple scales, of which process commonly involves several datasets obtained from different sensors. This paper presents a weighted least square based multi-sensor data fusion method for such measurement. The method starts from unifying the coordinate frames of the measured datasets using an intrinsic feature based surface registration method. B-spline surface is used to fit linear surface model to each identified overlapping area of the registered datasets, respectively. By forming a common basis function, the fitted surface models and the corresponding residuals are then combined to construct a weighted least square based data fusion system which is used to generate a fused surface model. An analysis of the uncertainty propagation in data fusion process is also given. Both computer simulation and actual measurement on various freeform surfaces are conducted to verify the validity of proposed method. The results indicate that the proposed method is capable of fusing multi-sensor measured datasets with notable reduction of the measurement uncertainty. (C) 2016 Published by Elsevier Inc.