NSGA-II approach for proper choice of nodes and knots in B-spline curve interpolation

Abstract

Describing data, obtained by various instruments, with an analytic function is one of the tasks that people often face in a wide variety of applications such as virtual reality, CAD design, reverse engineering, data visualization, medical imaging, and cultural relic restoration and so on. Moreover, non-uniform B-spline is an extensively-used tool for interpolation which is an effective means of describing data. In this paper, according to the geometric features reflected in the data points, a method for calculating the tangent vectors at the corresponding data points is proposed for reference. And based on the constraints on tangent vectors (calculated by the proposed reference method, can also be given beforehand) and control points, non-dominated sorting genetic algorithms-II, namely NSGA-II, is adopted for adaptive B-spline curve interpolation without knowing nodes and knots in advance. The resulting interpolation curve approximates the given tangent vectors and the data polyline, and it is more natural-looking, in general than those obtained by other methods. In addition, the new method works well in a higher degree. Testing results on the feasibility and universal applicability of the new method are also included.