Improved Filtering and Hole Filling Algorithm for the Point Cloud of Rotor Surface Based on PCL

The quality of rotor is key to ensuring the quality and stability of the blower. Point cloud-based detection utilizes three-dimensional depth information, making it time-saving, efficient, and particularly suitable for rotor quality inspection. In order to ensure efficient and accurate detection, the collected rotor point cloud needs to be preprocessed. The preprocessing includes filtering and hole filling. Therefore, to tackle the problem of outliers and isolated points in the point cloud on the rotor surface, this paper proposes a straight-through filtering algorithm based on region selection. Our method solves the problem of inadaptability for traditional straight-through filtering to curved point clouds, and ensures the speed and elimination effect. In addition, to address the problem of holes in the rotor point cloud, this paper proposes a hole-filling algorithm based on neighborhood statistics, which compensates for the misjudgment of hole positioning in the hole-filling algorithm based on cubic spline interpolation, and improves the effect of hole-filling. Experiments show that the combined application of the two proposed algorithms provides a relatively complete and smooth rotor point cloud for quality inspection, thereby guaranteeing the quality and robustness of the rotor.