Scale-invariant Road Sign Recognition By Fourier Plane Scaling Of Spectrum

摘要

A new method is proposed for estimation of object size from the Fourier spectrum leading to scale-invariant pattern recognition. Instead of using the standard procedure of radial harmonic functions, the scaling factor is estimatedfrom the Fourier spectrums of reference and test images. Since Fourier Transform of a two-dimensional object and its Fraunhofer diffraction pattern are identical, it is utilized to determine the scale factor. The test image is correlated with the reference image after reseating its Fourier spectrum. Computer simulation results are presented for a reference object at several scaling versions as well as for some non-target objects.