A new quadratic and biquadratic algorithm for curve and surface estimation (Q1315795)

The author presents a new algorithm for curve and surface approximation. The algorithm does not assume that the surface to be estimated and based on a given set of data in the three-dimensional space has a continuous first derivative, nor does it assume that the data satisfy the assumption of stationarity or the intrinsec hypothesis. The algorithm is faster then splines and kriging. Four functions are presented in various methods. Rich illustrations present the estimation of each function and the estimated error surface. A comparative analysis with bicubic natural splines based on data generated from known functions is also given. The speed of execution recommends this algorithm for real-time applications.