Improvement of the performance of optical correlator by using fuzzy theory

摘要

In this paper, based on fuzzy theory, we had discussed the principle and technique of gray-tone image recognition and proposed a novel template-matching structure using optical correlator. In this novel method, a fuzzy relationship matrix, which was constructed by triangle fuzzy set, had been defined to describe the characteristics of gray-tone pixel matching. A changeable m-bit cycle-encoding method has been introduced to realize the fuzzy relation pixel-matching on a correlator structure. We also discussed how to select m correctly for different distortion in gray-level object recognition. This new method improved the performance of the usual optical correlators and got the best balance between the spatial-band width and the processing accuracy. According to this method the optical experimental results are obtained by using an incoherent correlator. We had built up a compact incoherent correlator for gray-tone object recognition.