A calibration method based on outlier-removal for high-accuracy uniaxial MEMS-based reconstruction

High-accuracy calibration of uniaxial MEMS-based systems is essential for achieving precise 3D reconstructions because the calibration results will directly affect the accuracy of 3D reconstruction. Calibration data consists of two parts: absolute phase and corresponding 3D coordinates. However, due to the unstable motion of MEMS scanning mirrors and defects in the calibration board, the obtained calibration data may contain errors. These erroneous data points can significantly reduce the calibration accuracy of MEMS systems. To address this challenge, we propose a high-accuracy global calibration method based on an outlier removal strategy. First, the proposed method obtains calibration data using a calibration board with a checkerboard or dot matrix pattern. Then, the system is calibrated using the improved RANSAC algorithm. During each iteration, the fitting residual for each calibration data point is calculated based on the calibrated model parameters. The optimal calibration model is iterated based on minimizing the total fitting residuals. In this process, outliers are removed as they would lead to significant calibration errors. Experimental results show that the proposed method effectively mitigates the impact of negative factors on system calibration, such as the unstable motion of the scanning mirror. The outlier removal mechanism enhances the accuracy of system calibration, thereby achieving accurate uniaxial MEMS-based reconstruction.