An optical object recognition system using log-polar coordinate transform of power spectrum and NJTC

파워스펙트럼의 Log-polar 좌표변환 및 NJTC를 이용한 광 물체 인식 시스템

In this paper, we propose a new opto-digital object recognition system which has rotation, scale, and shift invariant characteristics. The fourier power spectrum of the object image is modified to get shift invariance. The log-polar transform is used for rotation and scale invariance. And the decision of similarities is performed by nonlinear joint transform correlator (NJTC) that can control the ratio of phase and amplitude signals. Experimental verification of th eproposed optical object recognition system is presented.