A method for improving coordinate transformation accuracy of multi-station visual measurement systems in large-scale measurement

Abstract The multi-station visual measurement (MSVM) system is widely applied in the measurement of large-scale complex components. The local complex structures of the component can be accurately measured by multiple stations of the local visual system (LVS). However, the overall measurement accuracy of the MSVM system is limited by the accuracy of coordinate transformations between LVSs. This paper proposes a method to improve the coordinate transformation accuracy of the MSVM system. Firstly, an uncertainty evaluation model is built for the global coordinate system (GCS) constructed by the multilateration method. Secondly, a comprehensive error model for LVS measurements is developed by integrating the characteristics of the systematic error and the random error. Additionally, an orthogonality-constrained weighted total least squares method is proposed to improve the accuracy of coordinate transformation calculations, with its weights matrix determined from the built error models of the GCS and LVS. Finally, simulations and experiments validate the improvement in coordinate transformation accuracy by the proposed method.