Algorithm of scattered point cloud data reduction based on non-uniform subdivision

For reduction of scattered point cloud data, one algorithm based on non-uniform subdivision was put forward. The space partition of point cloud was generated using octree structure. k neighborhood was constructed through partition result. All the points in the k neighborhood were approximated by quadratic parametric surface of which the mean curvature determined whether to carry out non-uniform subdivision for the octree space or not. In the process of subdivision, the degree of subdivision depended on the maximum interval angle of each data point. Boundary points were identified and protected by constructing curvature difference function. The algorithm was applied for reduction of point cloud with curvature diversification. The reliability and accuracy of the algorithm were validated by experimentation.