A new method of fitting conic to scattered data

Using the implicit conic to fit the given data points is an important problem in the fields of geometric modeling and numerical calculation. If the implicit conic is constructed by minimizing the algebraic distance, the selection of the coordinate system and the setting of the constraint are the two key issues that will affect the approximation accuracy heavily. This paper discusses the problems that how to select the coordinate system and the constraint to make the constructed implicit conic approach the given data points better. Many experiments also show that the coordinate system and the constraint determined by the new method improve the conic approximation accuracy greatly.