Filter of point cloud data is a hot and key step in processing LIDAR point cloud.
In implementing slop-distance filtering based on moving window, this paper
presented Biharmonic Spline Interpolation method and Radial Basis Function
Interpolation to filter the point cloud, which enriched the arithmetic and theory of
point cloud filtering.
During the process of laser scanning, objects blocking will cause point cloud
incomplete, and later filtering through the point cloud also make ground data have
many gaps or "holes." The Way to solve point cloud missing is to repair it，one option
is to do supplement measurements with general measure means, such as GPS RTK,
Total Station etc. Another option is to insert value to calculate by taking full
advantage of the point cloud around the "holes" and terrain similarity of the “holes”.
To insert value for dishevelled point cloud, besides Convex Delaunay Triangulation
linear interpolation, cubic interpolation, the nearest neighbor interpolation,
Biharmonic Spline Interpolation method (BSI) and radial base function (RBF) method
are introduced to interpolate the point cloud. On the other way, we present SVR as an
interpolation method to repair the point cloud. All of which enrich the interpolation
ways for point cloud data, this paper also submits a standard of evaluating
interpolation ways suitable for different terrain combining with related measure
criterion, not make a simple evaluation for interpolation form statistics.
Point cloud modeling is the core of point cloud processing. For TIN modeling to
point cloud, this paper adopts classic sub-governance arithmetic to configure network,
which achieved a good network configuration speed and quality, and also expounds
how to track the isoline and cut the section line according to TIN, Carries out a simple
tensile modeling for a house point cloud, and then makes the DSM model of point
cloud.
Based on the research results of above theories, we develop a point cloud
processing software PCEdit, which can be embedded in AutoCAD system as a
module.
Key Words：LIDAR，Point Cloud Filter，Point Cloud repairing， Biharmonic
Spline Interpolation，Radial Basis Function Interpolation