An Optical Pattern Classification Using Computer-Generated Holograms

We have proposed an optical pattern classification system which performs weighting and summation by the optical system using computer-generated holograms (CGHs). Our system makes it possible to simplify the structure and to improve the processing speed. We propose a trainable pattern classification system which can perform weight modification by using a CGH matrix and a dot matrix liquid crystal display (LCD) as a shutter array. We describe the trainable pattern classifier using a CGH matrix and LCD, and present the experimental results.