Investigation Of Matched Filtering Correlator Based On Photorefractive Volume Holographic Storages

The matched filtering correlator (MFC) performs two-dimensional operation capabilities with high speed and parallelism. There exist some advantages in photorefractive volume holographic storages, such as high spatial resolution and large storage capacity etc. The individual advantage can be employed by combining the volume storage with the MFC. The MFC of object images with various angles and orientations may be memorized in photorefractive crystal at one time. The matched filters with large template sample library makes correlation system take on real-time, multi-targets and distortion-invariant recognition capabilities. The key of the recognition system is on the reconstruction of matched filter. In order to obtain good recover filter, the channel encoding and equalization, denoising method are introduced into correlation system. For the correlation system, the principal works in the thesis are as follows:(1) The Wiener filter equalization, constrained least mean square equalization and CMA adaptive blind equalization are employed in MFC based on Photorefractive volume holographic storages. The simulation results show that the BER of reconstructing images after equalization is largely reduced and the images after equalization can yield an output BER of 10-3 when the SNR exceeds 30dB.(2) The solving of Blind equalization in practice is very difficult. For the conventional channel equalization techniques, the channel point-spread function (PSF) must be determined firstly, but it is almost impossible to obtain the PSF in practice. Hence an estimate method for Gauss model PSF based on the BER magnitude is presented. The simulation results under Wiener filter equalization show that the estimate method has rather good practical value only if the SNR of system is high enough.(3) The noise of volume holographic storage system is the superposition of multi-form noise, so the denoising method must be capable of eliminating mixed noise. By investigating various image denoising methods, a synthetic filtering denoising method based on LoG(Laplacian of Gaussian) method of edge detection is presented. The good effect is demonstrated in the simulation experiments.(4) The photorefractive volume holographic storages and matched filtering correlator are combined to form photorefractive volume holographic storage matched filtering correlator, and the channel encoding and equalization are introduced into this system. The simulation results indicate that the correlation system has rather good recognition performance.(5) In view that the distortion-invariant ability is an important capability of the opticalcorrelator, the correlation recognitions with five common distortions, scale distortion and rotation distortion are researched in detail by numerical simulations. The simulation results perform that the five common distortions can be controlled in a certain range and have no intense influence on correlation recognition, the rotation distortion brings more intense infection and the scale distortion brings rather intense infection on correlation recognition.(6) The MFC performance to suppress noise is investigated particularly. The simulation results indicate that optical correlator performs well in suppressing noise whether Gaussian or impulse noise as long as the SNR exceeds 20dB.