Local-to-global registration method of large complex components based on a de-pseudo-weighted variance minimization algorithm
The machining quality of complex components,such as the automobile flywheel shell and body,highly depends on measurement technology.Focusing on the problem of matching deviation caused by structural deviation,uneven margin distribution,and various inherent measurement defects,in this paper,a de-pseudo-weighted variance minimization algorithm is proposed for fine registration according to the idea of local-to-global registration.A weight function adjusted via an adaptive scale factor is established to distinguish the structural deviation point cloud from other abnormal point clouds.Thus,a reasonable weight is applied to the distance between each point,and the influence of the structural deviation point cloud on the registration accuracy can be effectively reduced.Furthermore,the adaptive coordination distance is established by the unified point-to-point distance and the point-to-plane distance to enhance algorithm convergence stability.The algorithm proposed in this paper can effectively enhance registration accuracy and inhibit the matching distortion in the process of local registration and global positioning of the automobile flywheel shell.Compared with the iterative closure point and variance minimization algorithms,the absolute positioning accuracy is increased by 18.9％and 66.7％,respectively,and the matching distortion inhibition degree is improved by 25.9％and 85.3％,respectively.Additionally,compared with the weighted plus-and-minus allowance variance minimization algorithm,the convergence stability is improved.Moreover,the proposed method can effectively register the positioning with only a single data acquisition,greatly improving registration efficiency.
万方数据
维普
