作者: Chen Zhang , Yong Lei , Linmiao Zhang , Nan Chen
DOI: 10.1080/24725854.2017.1348646
关键词: Cartesian coordinate system 、 Algorithm 、 Surface reconstruction 、 Simulation 、 Gaussian process 、 Laser scanning 、 Surface (mathematics) 、 Regression 、 Engineering 、 Derivative 、 Taylor series
摘要: ABSTRACTThis article presents a Gaussian process (GP)-based approach to model tunnel’s inner surface profile with high frequency sensing data provided by Terrestrial Laser Scanner (TLS). We introduce reading-surface that uniquely determines three-dimensional tunnel in Cartesian coordinate system. This transforms the cylindrical two-dimensional profile, hence allowing us GP. To account for errors induced TLS, we take repeated measurements at designed coordinates. apply Taylor approximation extract mean and gradient estimations from then fit GP both obtain more robust reconstruction of profile. validate our method through numerical examples. The simulation results show help derivative estimations, outperforms conventional regression noisy observations terms mean-squared prediction e...