Abstract: A new approach is proposed to recognize aircraft automatically. To describe the characters of aircraft simply and efficiently,moment invariants and Fourier descriptors are combined. A neural network with distributive storage ability is used. At the same time, several methods are taken to accelerate its convergent process and experiments of nine different aircraft have proved that this recognition system performs well; it has an accuracy over 99% and high recognition speed. The whole recognition process can be completed within 1 second.

Key words: moment invariant, Fourier descrip to r, BP neural netwo rk s