Research On Preprocessing Technology Of Reverse Engineering Point Cloud Data

With the 3D point cloud data acquisition technology continues to develop,making the physical surface of the digital has become increasingly simple,while precision is also increasing.However,no matter what kind of data acquisition method will be affected by human factors,environment and other factors,the inevitable existence of measurement errors,the existence of these errors will directly affect the quality of three-dimensional reconstruction model.At the same time with the improvement of measurement accuracy,the number of data points collected on the surface of the object is also very large,in the point cloud data processing,storage will seriously affect the efficiency of three-dimensional reconstruction.Therefore,it is very important to study the 3D point cloud data processing technology.In this paper,we mainly study three-dimensional scattered point cloud data processing technology,including k-nearest neighbor search algorithm of point cloud data,point cloud data denoising and data reduction,the main contents of the paper are as follows:In the space division of the point cloud data,the calculation method of the minimum grid edge length of the bounding box method is improved,so that the search efficiency of the k nearest neighbors is obviously improved.The minimum bounding box of point cloud data is established,and the first spatial division of the data points in the bounding box is given by the given grid edge length,so the "black body" occupancy rate can be calculated.After the first division,The distribution of point clouds and the black body occupancy rate are considered synthetically,which makes the division of the lattice variable length more reasonable for different point cloud data,and accelerates the speed of k nearest neighbor search.The outliers and noise points in point cloud data affect the quality of 3D point cloud data reconstruction.In order to eliminate these noise while preserving the geometrical features of object surface,a point cloud data based on normal vector correction is proposed.Noise Smoothing Algorithm.The outliers in the point cloud data are filtered out by using the statistical method.The geometric features of the point clouds are estimated by the weighted covariance matrix of the noise points.The points with similar geometric characteristics are restricted to regions where the normal vectors are similar.The normal vector and position of the sampling points in the neighborhood are guaranteed to be trilinear.The algorithm can not only filter out the large noise signal,but also keep the sharpness and edge characteristics of the point cloud data model while smoothing the small noise.An improved algorithm of the simplification of the three-dimensional point cloud data is described.A algorithm of simplification combining curvature and normal vector is proposed.This method can avoid the feature information loss of three-dimensional point cloud data where the curvature of the surface no significant change but it is a feature information,so surface feature information can be well preserved by using the method.