Using cross-ratios to model curve data for aircraft recognition

This study proposes a novel method that utilizes the projective invariance of NURBS (nonuniform rational B-splines) and cross-ratios to recognize aircraft in images. An aircraft contour extracted from an image can be fitted by a NURBS curve in approximation, and a skew-symmetry detection method is employed to determine the start point of the contour for the fitting curve. The control points on the fitting curve are then utilized to calculate a set of cross-ratios to represent the aircraft contour, and a small database is established for aircraft recognition. Moreover, all real cross-ratios are converted to integral cross-ratios to increase the error tolerance and matching facilitation. The experimental results demonstrate that the proposed method is robust and effective in aircraft recognition. (C) 2003 Elsevier Science B.V. All rights reserved.