Author Affiliations
Abstract
State Key Laboratory of Precision Measurement Technology and Instruments, Department of Precision Instruments, Tsinghua University, Beijing 100084
A volume holographic wavelet correlation processor for performing rotation invariant pattern recognition is suggested. It uses wavelet transform to get the digital edge extraction of the original object and a single circular harmonic component is used as the matched filter to get good rotation invariance. The new filter used in this method is called wavelet circular harmonic component filter (WCHCF). Simulation results validate the theory and the experiment to recognize human faces with any rotation angle shows the utility of the newly proposed method.
100.5010 pattern recognition and feature extraction 350.2460 filters interference Chinese Optics Letters
2003, 1(2): 0274