Surface fitting to scattered data by a sum of Gaussians (Q1803688): Difference between revisions

The authors report on numerical experiments with interpolation and approximation by a sum of Gaussians (SOG). They study how the parameter \(\sigma\) of the Gaussians influences the accuracy of the computation and the accuracy of the interpolation of a function. They describe an algorithm for the approximation of scattered data by an SOG such that the root mean square error becomes small. The first estimate of the parameters is obtained by the Hough transform. The parameters are then optimized by the Marquardt algorithm.