基于径向基函数和自适应单元分解的大规模散乱点云快速重构

Fast Reconstruction of Large Scattered Point Clouds Using Adaptive Partition of Unity and Radial Basis Functions

径向基函数用于散乱数据的插值和拟合具有精确和稳定的优点,但是它不适合大规模点集的曲面重构。把径向基函数和单元分解原理综合起来,提出一种大规模散乱点云的隐式曲面快速重构算法。把整体定义域自适应细分成一系列稍微重叠的子域,基于径向基函数在各子域上计算局部表面,最后采用单元分解函数对局部表面进行加权混合得到全局的重构表面。方法适于处理数量较大和分布密度变化较大的点云数据重构。

Radial basis functions are accurate and stable for scattered data interpolation and approximation, but they are not appropriate for reconstructing surfaces from large point sets. An efficient algorithm for implicit surface reconstruction for large-scale scattered data was proposed based on the combination of radial basis functions and partition of unity. The whole domains were adaptively subdivided into a set of slightly overlapping subdomains, in which smooth local surfaces were then computed using radial basis functions. Finally the whole surfaces were reconstructed by weighted blending of local surfaces with partition of unity functions. The algorithm is robust for the reconstruction of large point cloud data with varying density.