Improved Robust Hand-Eye Calibration Algorithm Based on Data Optimization and Filtering

The mathematical model of hand-eye calibration is an important model in machine vision area. But in many other areas, the mathematical model of hand-eye calibration can be also used to get the transformation relationship of two frames, such as computer-aided diagnosis. The quality of input data is an important factor for the hand-eye calibration algorithm to obtain accurate result. Even if the algorithm is excellent, the unsatisfactory data will have a significant influence on the accuracy of the algorithm. Therefore, a data filtering rule is employed to eliminate the outliers in data sets in this paper. Firstly, an initial result is obtained by the robust hand-eye calibration algorithm using dual quaternion. Then, the outliers are filtered by using the data filtering rule. In this paper, we use l
2
matrix norm to quantify the error values that are obtained by AX-XB. By sorting the errors from small to large, the threshold is set to filter the data with large error values. Finally, the filtered data is used to calculate the new result by the robust hand-eye calibration algorithm. We implement the backprojection experiments to verify the validity of the proposed method. The proposed method is more accurate and it has a better robustness. Compared with the robust hand-eye calibration algorithm with dual quaternions, the accuracy of the result is improved by 50% after filtering the data.