The Research On Hole Repairing For 3D Laser Scanning Point Cloud

In the process of 3D laser scanning data acquisition,it is inevitable to be affected by the factors such as occlusion,the limitation of scanning method and the complexity of the surface shape of the object to be measured,which will cause the missing holes in the data.In order to ensure the complete modeling of point cloud data by subsequent reverse engineering,it is necessary to carry out filling for the missing part of point cloud data.In this paper,the corresponding repair algorithm is realized by MATLAB programming,and then 3Dreshaper software is used to complete the visualization of the point cloud model and compare the repair effects of different algorithms.The main contents include the detection of hole boundary points based on dimension-reduction method,the acquisition of hole boundary edges and the filling of hole based on cubic B-spline curve fitting method.The main work and achievements are as follows:(1)The platform carried by the 3D laser system and the acquisition mode are divided,and the differences between them are analyzed in detail.At the same time,the types and pretreatment methods of point cloud data are introduced in detail,which lays a foundation for the subsequent point cloud processing work.(2)Since the data obtained from the initial scan does not contain any topology information,the point cloud neighborhood construction is required in the first place.After a detailed analysis of the advantages and disadvantages of several common construction methods,this paper uses the construction method based on KD tree to find the k-near point of the point cloud.At the same time,it also analyzes the estimation method of the normal vector of the local geometric attribute of the point cloud.Principal component analysis(PCA)method is used to construct the covariance matrix based on the K near point to estimate the normal vector and adjust the normal vector direction through the view point direction.(3)Aiming at the problem that the traditional four-quadrant method cannot accurately identify the boundary points of holes when there are points in four regions,this paper proposes a method combining the four-quadrant method with the maximum Angle criterion on the basis of existing methods.In this algorithm,the idea of dimensionality reduction is adopted to extract the boundary points from the target point cloud data and connect them as boundary lines to complete the identification of hole regions in the point cloud model.The experimental comparison is completed by programming,and the results show that the algorithm in this paper can correctly identify the holes in the point cloud model when the appropriate threshold is selected.(4)In order to get a better repair effect of the hole area,the cubic B-spline algorithm was used to fit the missing area of the point cloud,and the hole repair experiment was completed.Moreover,the proposed method is compared with common repair algorithms such as those based on radial basis function and least squares surface fitting.The proposed algorithm achieves relatively better repair effect in the above comparison.The experimental results show that the proposed algorithm is easy to implement and the fitting surface transition is natural.The desired repair effect is achieved and the algorithm has good practical value.