Abstract: Aiming at the problem that the feature points in point cloud data contain gross errors, a robust point cloud registration method based on dual quaternion is proposed. The robust point cloud registration model based on dual quaternion is constructed according to the principle of weighted least squares. The solution formulas of rotation matrix and translation vector are derived by Lagrange multiplier method. The iterative steps of new weight calculation and coordinate transformation are studied. The example shows that the gross error can be effectively identified and eliminated with the method, and the accuracy of point cloud registration is improved.

Key words: dual quaternion, weighted iteration method, Lagrange multiplier method, point cloud registration, gross error