Gradient Vector Estimation Using Quadratic Regression Function

Abstract

An extension of 4D linear regression method for scalar iregularly distributed volumetric data gradient estimation will be presented here. The aim is to reach higher accuracy and the main tool is using quadratic regression function. The results will be compared to the original method as well as to the approach based on generalization of the finite differences method. The performance of all the three method will be examined from different points of view.