Study On Uniformality Of Correlation Peaks And Miniaturization Of Volume Holographic Correlation System

Volume holographic correlation technology is a potential optical pattern recognition technology that can be used to deal with large batch of image data. In this dissertation, holograms are stored in doped LiNbO3 single crystals, and a miniaturized volume holographic correlation system is developed.Based on the theory of coupled wave, we study the influence of diffraction enhanced effect on the diffraction efficiency for image storage in LiNbO3 by angular-multiplexing technology. Combining the influence of angular, we can analyze the diffraction efficiency of each stored pattern. Aiming at pattern's recognition, the exposure time is set to be rather small, consequenly, under somen special experimental paramenters; storage of equal diffraction can be realized with the influence of erasing effect and angular simultaneously. Experimentally, arrays of correlation spots of equal diffraction efficiency are obtained under the same exposure time.In order to make the recogniton system be high efficient and stable, a controlling system is designed. So, experimental paramenters, such as multiplexing angle, exposure time, image shuttering time, delaying time for platform shifting, can all be set automatically.Volume holographic correlation recognition technology is based on the intensity of each correlation peaks. But an accurate recognition directly basing on the intensity of correlation peaks is very hard. In this thesis, pattern recogniton and dynamic pattern recognition methods are used.A miniaturized correlation recognition system of 350×200×70mm~3 manufactured in the present work, which includes transmission and reflection geometry system. 2010 patterns are recognized successfully at a location of storage medium. In order to increase the life time of stored patterns, thermal fixing technology is developed to make those stored patterns non-destructive during the recognition processes.By adding a rotating in the developed miniaturized recogniton sysutem, 10000 patterns are stored. Furthermore, a real-time image-collecting system is designed and added in the recogniton system. Real-time recognition of 25 human face pictures is implemented preliminarily. To broaden the application fields of the current developed system, concave lens and focusing lens are introduced in and the recognition of 3D object is achieved initially.Conventional techniques can not record and erase patterns dynamically, because all the stored patterns (or holograms) overlap in the same volume of the recording medium. In the recording medium, a stored pattern is affected by the optical beams required for recording or erasing. In this dissertation, local holographic correlation was investigated by storing images in slices of a crystal; erasing and recording patterns dynamically can be realized.