Terrain Model Reconstruction from Terrestrial LiDAR Data Using Radial Basis Functions

Abstract

The presence of vegetation and the terrain topography itself generate strong occlusions causing large gaps in terrestrial laser scanning (TLS) data at the ground level as well as a risk of integrating above-ground objects. This article introduces a surface-approximation algorithm dedicated to extracting digital terrain models (DTMs) from terrestrial TLS data acquired in forest areas. The proposed method is based on the combination of a quadtree subdivision of space guided by the local density and distribution of data together with a surface modeling via radial basis functions, which are used as partitions of unity for merging local quadratic approximating patches.

Harmonic current classification using hybrid FAM-RBF neural network

“…On the other hand, the RBF has been catching the attention of researchers recently with the function approximation of its hidden neurons [17]. This research includes enhance the performance of RBF and apply RBF in various industries [7,[18][19][20][21]. RBF parks under the group of feedforward NN, which organizes its hidden neurons with function approximation theory.…”

Calibration Method of Orthogonally Splitting Imaging Pose Sensor Based on KDFcmPUM

“…RBF interpolation is used to display local features in the sub-domains, then local features are restored globally by a weight function. This method is widely used in image segmentation [16,17,18] and three-dimensional surface reconstruction [19,20,21]. Different methods have been proposed to partition the domain and construct the weight function to improve the PUM’s management of large scattered data.…”