Evolutionary heuristic optimization for digital curves and surfaces

Abstract

Computing curve and surface fitting is the main problem encountered to approximate data in many graphics and image processing applications. For approximation of complex curve and surface fitting, NURBS (Non Uniform Rational B-Spline) is the most popular flexible spline technique. Control points, weight vector and knot vector are the NURBS parameters which effect shape of curve/surface. For different values of these parameters we get different shapes of curves/surfaces. This research is concentrated to a common area among the disciplines of computer Graphics, Imaging, and Vision. The idea is mainly to optimize the contours obtained from the outline of images. In this research we have applied simulated Evolution (SE) for optimization of control points, weight and knot parameters of NURBS for curve and surface fitting. The main objective is the reduction of fitting error with target curve/surface to obtain a smooth curve/surface in a reasonable time. (Abstract shortened by UMI.)