基于RBF神经网络的点云滤波与空洞修复研究

三维激光扫描仪作为一种新型高科技产品,它的应用已经渗透到国民经济的各个方面。如何高效地对点云数据进行滤波以及空洞修复,已成为当下研究的热点问题。针对目前点云滤波与空洞修复中存在的效率与准确性等问题,利用RBF(Radial Basis Function)神经网络最佳非线性逼近以及快速收敛能力,提出了一种基于RBF神经网络的点云滤波与空洞修复算法研究。通过真实扫描数据进行实验,结果显示该算法具有很高的预测精度,并且对点云空洞具有很好的修复效果,可为实际工程应用提供参考。

Research on Point Cloud Filtering and Cavity Restoration Based on RBF Neural Network

As a new high-tech product,three-dimensional laser scanner has penetrated into all aspects of the national economy. How to filter the point cloud data efficiently and cavity restoration has become a hot issue in current research. Aiming at the problems of efficiency and accuracy in point cloud filtering and h cavity restoration,a point cloud filtering and hole repairing algorithm based on RBF neural network is proposed by using the best nonlinear approximation and fast convergence ability of RBF neural network. Experiments on real scanning data show that the algorithm has high prediction accuracy and good restoration effect on point cloud cavities,which can provide reference for practical engineering applications.