Pattern classification using optical Mellin transform and circular photodiode array

Scale-invariant pattern classification using a hybrid system combining the optical Mellin transform and a digital signal processing technique is discussed. We accomplish the optical Mellin transform by a logarithmic coordinate transformation using a computer-generated hologram, followed by an optical Fourier transform. Mellin transform patterns are detected with a circular photodiode array, whose output signals are processed by a micro-computer. A new criterion is discussed, in which circular or periodic correlation is employed. Experimental examples are presented.