Denoising of a Multi-Station Point Cloud and 3D Modeling Accuracy for Substation Equipment Based on Statistical Outlier Removal

In order to construct a three-dimensional model of substations, we need to use laser scanning to collect the information of substation equipment from different angles and construct a three-dimensional model of substation equipment by splicing the acquired point cloud data. Because there are many noise points in the point cloud data of substation equipment acquired by laser scanning, it directly affects the modeling accuracy of the three-dimensional model of substation equipment. Aiming at investigating the characteristics of the high density of point clouds, the low density of noise points, and the sparse distribution of substation equipment, based on the Statistical Outlier Removal (SOR), this paper focuses on the influence of the standard deviation of the kernel function and the number of points estimated within the average distance on the noise denoising effect of a point cloud. Firstly, the binary approximation method is used to optimize the threshold value of the kernel function for the point cloud data of the substation equipment obtained under laboratory conditions, and the optimal threshold value of the kernel function is determined according to the evaluation of the noise denoising effect of the point cloud. After the optimal threshold noise denoising, the multi-site cloud of substation equipment is spliced, and the influence of the noise denoising effect on the accuracy of multisite cloud splicing is analyzed. Finally, the obtained optimal threshold of the kernel function is used for noise denoising and multi-site cloud splicing of point cloud data from the same kind of substation equipment collected in an actual environment, and the noise denoising effect and the splicing accuracy of a three-dimensional point cloud of the same kind of substation equipment under the optimal threshold are tested. The field application results show that the optimal threshold obtained by this paper for similar equipment can effectively improve the efficiency of point cloud noise denoising and the accuracy of multi-site cloud splicing, and verify its effectiveness and correctness. This method can improve the efficiency of digital acquisition and 3D modeling, and reduce the cost of 3D modeling.