Improvement of Fourier descriptor using spatial normalization

This paper presents a novel normalization for the Fourier descriptor, named as spatial normalization Fourier descriptor, which presents a new method to normalization the Fourier descriptor. Especially in the finding start point there is innovation comparing with former algorithm. This invent greatly improve the performance of the Fourier descriptor, and provides high robust and invariant. The same time its foundation in theory is obvious and that make it easy to be understood and fulfilled. It will greatly impulse the pattern recognition of shape, and in the so many applications for the Fourier descriptor it is highlighted. Lastly the experiment verifies it and it outperform the traditional Fourier descriptor.