Scale-invariant three-dimensional object recognition using a classical convergent correlator

摘要

We present an optical method to obtain scale-invariant three-dimensional object recognition. It is based on using the Fourier transform profilometry technique and a classical convergent correlator. To obtain the scale invariance property different approaches, such as Mellin radial harmonic decomposition, logarithmic radial harmonic decomposition, and specifically modified matched filters, are considered. Simulated results demonstrate the good performance of the method. Scale-invariance is achieved in an interval of scales 1:1.67.