A Hybrid Interpolation for Reconstructing Near-Surface Model

When both the number of control points and the size of grid become large,TPS based reconstruction of near-surface model is often very time-consuming.Therefore,such method affects the efficiency of building near-surface model in the static correction.To address this problem,a hybrid interpolation combining TPS with cubic spline interpolation is used for reconstructing near surface model.Firstly,not only recursive LU decomposition but also GPU based LU decomposition is used to solve large linear system of equations.And then,TPS interpolation function is created.Secondly,the grid is rarefied with appropriate steps in the X and Y directions and then,TPS interpolation function is evaluated on the sparse grid.Based on which cubic spline interpolation function is created and then,it calculates the value of the other points.Finally,OpenGL visualizes the 3D near-surface model.The experimental results show that this algorithm speeds up the reconstruction of near-surface model and approximates the TPS interpolation in accuracy.