Neural network based fairing of digitized curves and its application

Abstract

A fairing method for digitized curves, based on neural network optimization, is presented in this paper. The energy function and constraints of a digitized curve are given first, with which the fairing process is converted into a quadratic optimization problem. A neural network is then used to solve the optimization problem. Fairness adjustment of points' position is realized by this optimization. An application to binary image magnification is illustrated, which shows better result than other methods. © Springer-Verlag Berlin Heidelberg 2005.