Genetic fourier descriptor for the detection of rotational symmetry

In this paper, a Genetic Fourier Descriptors is proposed to detect rotational symmetry. Rotational symmetry is one of the important features for image decoding and object recognition in computer vision systems. In the Genetic Fourier algorithm, the Fourier descriptors are chromosomes and fitting function of the GA. The Genetic Fourier method has the following advantages. (1) It can handle partially occurred contour and opened contour, (2) It can handle complex point pattern, (3) It can obtain multiple perceptions and (4) It is highly parallel and its efficient can be greatly improved if parallel processors are used. Experimental results show that it can handle complex symmetry figures, these symmetry figures may be formed by separated curves, points or partially occurred or partially missed (open contour). Copyright © 2006 Springer-Verlag Berlin Heidelberg.