Artificial Neural Networks (ANNs) are usually designed around vector-matrix multipliers, where the inputs to the neurons are represented by the vectors while the interconnection weights are represented by the matrix. Optics, with its interference-less free-space communication capabilities, is therefore an efficient and natural way to implement ANNs; however, it is not without practical problems. In electro-optic ANNs, errors due to non-linear or limited accuracy components could exist in the input devices, the interconnection weight matrix, or the output detectors. This report addresses some of those errors in terms of a specific implementation and across several ANN architectures. The electro-optic software only. In the electro-optic layers, light emitting diodes (LEDs) are used to provide the input, liquid crystal spatial light modulators (SLM) serve as the interconnection weight matrixes, and photodiode detectors are the nonlinear thresholding elements. Specific hardware imperfections - nonuniform LED illumination, optical misalignment and cross talk within the SLM, thermal drift in the SLM, and noise and linearity problems in the photodiode detectors circuits - are analyzed and experimentally documented. The impact of these errors on the performance of an ANN is dependent upon the ANN architecture. These error sources, as they effect the design of this or any other electro-optic ANN, are then discussed and evaluated for several representative ANN architectures. Many decisions must be made when designing a practical implementation of an electro-optic ANN. This work provides some basis of how such decisions may be made.

Emission or propagation of optical irradiation through cylindrical objects can be described by the Abel equation. Only some of the numeric solutions have satisfactory accuracy. They provide convergence of outcome function but greatly increase calculation time. One of the ways to increase the speed of calculation is by optical processing of signal. Ultrafast processing never before realized with conventional electronics is available in decision of inverse problems. In the study we use optical derivation of a signal. Derivation of phase is easy in interference tomography. Mathematical algorithms became faster.

An analysis to estimate the effect of transducer apodization and discontinuities in acousto-optic devices is presented. Theoretical and experimental results show that a transducer sectioned with oblique cuts does not increase side-lobe levels while cuts normal to the optical propagation direction produce objectionable side-lobes far from the central peak.

Acousto-optic tunable filters have ben well documented throughout the last 30 years. Unfortunately, although the filtering characteristics of these system have been explored until recently little has been done with regards to the data analysis or exploring the advantages of processing multi- spectral information. Here we use the wavelet transform to process multiple banded images like ones acquired in an acousto optical system. We demonstrate the uses of spectrally sensitive wavelets to detect a small object presents by suing an arbitrary low-resolution image in a high clutter background.

A novel approach for 1D and 2D acousto-optic deflector design is presented. 2D deflectors are optimally built of two adjacent Bragg cells in acoustically rotated configuration. The depleting effect of second order diffraction is compensated by a small change of the optical incidence angle. High bandwidth devices working below 100 MHz acoustic frequency and 1 W acoustic power are reported. Several experimental results on deflectors for 633 nm and 1064 nm optical wavelengths are presented.

A robust signal processor, capable of handling a multitude of signal processing functions over a wide instantaneous signal bandwidth is needed for future military systems where shared sensor and signal processing resources will be employed. We investigated a novel 2D acousto-optic (AO) processor capable of real-time multifunction signal processing. We developed the specifications for an optoelectronic ISAR image formation processor for ship imaging based upon the AN/APS-137(H) Airborne Radar. The baseline processor is designed for high resolution imaging- slant range resolution of 0.3 m and cross range resolution of 1.2 m over 30 m X 30 m window. The optical design of the processor is based on an optically efficient, in-line, high stability, 2D interferometer using four acousto-optic devices invested by Riza. We developed specifications for the processor components, including light source, lenses, photodetector array, and Bragg cells including a multichannel Bragg cell to improve processor bandwidth and reduce its size. We showed that all of these components are commercially available. We breadboarded a narrow bandwidth version of the processor and demonstrated its main operation features. We showed experimentally that the processor has simultaneous spatial carrier generation capability that is controllable with high phase stability and excellent fringe visibility.

An electro-optic microwave signal processor for activity monitoring in an electronic warfare receiver, offering wideband operation, parallel output in real time and 100 percent probability of intercept is presented, along with results from a prototype system. Requirements on electronic warfare receiver system are demanding, because they have to defect and identify potential threats across a large frequency bandwidth and in the high pulse density expected of the battlefield environment. A technique of processing signals across a wide bandwidth is to use a channelizer in the receiver front-end, in order to produce a number of narrow band outputs that can be individually processed. In the presented signal processor, received microwave signals ar unconverted onto an optical carrier using an electro- optic modulator and then spatially separated into a series of spots. The position and intensity of the spots is determined by the received signal(s) frequency and strength. Finally a photodiode array can be used for fast parallel data readout. Thus the signal processor output is fully channelized according to frequency. A prototype signal processor has been constructed, which can process microwave frequencies from 500MHz to 8GHz. A standard telecommunications electro-optic intensity modulator with a 3dB bandwidth of approximately 2.5GHz provides frequency upconversion. Readout is achieved using either a near IR camera or a 16 element linear photodiode array.

The location of certain types of communication and radar RF emitters is a critical concern for a variety of purposes relating to the national defense, particularly for tactical communications and the detection and avoidance of threat radars. Determination of the angel of arrival of an RF signal is currently accomplished by processing the signals from a 2D array of sensor elements. The signals are typically processed with the use of an RF combination network and the performance is severely limited by interference and multiple signals of interest. This paper describes an acousto-optic signal processing architecture that provides the angle of arrival unambiguously over 360 degrees. The system architecture is based on a circularly disposed antenna array. The optical system that processes the antenna element outputs is based on a new A-O device design concept presented in this report. The operational properties of the new Bragg cell concept are illustrated.

A novel 2D acousto-optic processor is introduced for a variety of 2D signal processing tasks that include ambiguity function and range-doppler processing, 2D raster format high resolution signal spectrum analysis, and triple correlation function generation. The optical design is based on an optically efficient, in-line, high stability, 2D interferometer using four acousto-optic devices. Via the simple technique of Bragg cell carrier offset adjustments, the processor can generate the desired interferometric output on a chosen spatial carrier that is required for bias removal and electronic post-processing. The processor has bandwidth limitations based on the type of Bragg cells used in the system.

We present a multiple-input, single-output, weakly nonlinear model of a liquid crystal light valve using a second-order Volterra series and describe an experimental method to measure the nonlinear transfer functions using sinusoidal perturbation and synchronous detection with a lock-in amplifier. Experimentally measured and estimated nonlinear transfer functions are presented. We next discuss the response of the liquid crystal light valve to random inputs and present experimental noise measurements.

The use of polarization selective holograms for the implementation of photonic delay lines (PDL) is proposed. A single bit PDL using FLC devices as active polarization switches and polymer dispersed liquid crystal (PDLC) devices as polarization selective optical path routing components is experimentally demonstrated and characterized. Different within-channel leakage noise filters are discussed and experimentally demonstrated, and high signal-to-leakage noise ratios are obtained for both PDL settings.

Given are the results of experimental study on the quasi real time holographic correction for the lens distortions in the passive observational telescope in the visible range of spectrum, using the liquid crystal optically addressed spatial light modulator.

The subjects of the study were spatial light modulators (SLM), comprised by polymer photoconductor (PC) or a-Si_1-x:C_x:H photoconductor, and ferroelectric liquid crystal. The polymer PC provides the highest available diffraction efficiency (DE) among all layers when the spatial frequency of the hologram exceeds approximately 100 lp/mm. At the same time the parameters of the a-Si_1-x:C_x:H PC layer can be varied across the wide range and thus the proper tradeoff between the spatial resolution and reversibly can be chosen.

Proposed is the novel method of dynamic nonlinear-optical correction for distortions in wide spectral band. The method is based on combining of the negative optical feedback correction and dynamic holography correction in the system, using optically addressed phase modulators.

A new Fourier optical architecture based on polarization multiplexing has been developed. In contrast to the customary scalar designs, the two orthogonal polarizations of the light wave are used to realize two independent transmission channels within the same volume. This result in a compact Fourier optical processor whose volume is reduced by 75 percent in comparison with the canonical 4-f system. Additionally the new architecture also for fully electronical switching between the two operation modes namely correlation and spectrum analysis. The paper describes the basic parameters which govern the design of miniaturized optical correlators and elaborates on an actual realization of a compact Fourier optical processor which employs polarization multiplexing.

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.

Multimedia applications require high capacity and data transfer rates from secondary storage. Volume holographic memories are attractive because they can satisfy these requirements in the near future. These memories use a 2D page-oriented data format and they can perform associative recall of the data stored in them which is an important operation in a database system. We have conducted a set of experiments to characterize the associative recall in a volume holographic database system set up in our laboratory. We show that associate recall is very feasible and can perform with binary and grey scale data or a combination of both. We present results from experiments in terms of three applications: a) database management, b) interactive video, and c) video indexing. We show that the search argument can be quite small, however problems with sufficient optical power arise when the search argument is too small. For this reason we have also conducted experiments with optical amplification of the output signal from an optical memory. These experiments also indicate that photorefractive amplifiers perform well in such a system and can be used to increase the associative recall signal levels.

This paper describes a concept for scanning linear data tracks such as those found on Optical Memory Cards (OMCs), and shows several mechanisms by which the concept may be implemented. The scanner is made up of a combination of uniformly rotating components. In contradiction to conventional OMC scan mechanisms, the concept we present has no reciprocating components. A scan mechanism is described which relays an image of a small area on an OMC through a pair of periscopes to a location where it may be viewed by a conventional Optical Pickup Unit (OPU). The periscopes are constrained to rotate in such a way as to cause the small illuminated area to translate along a data track on the OMC in a linear fashion. Thus, from the fixed point of view of the OPU, it appears that the optical data spots are moving linearly across its field of regard, while in actuality, neither the OPU nor the OMC move. Only the uniformly rotating scanner mechanisms moves, and in so doing, it creates the appearance of relative linear motion between the OMC and the OPU.

Some new techniques and architecture based on the concept of holographic interferometry with the use of photorefractive crystals as real-time recording and reconstruction devices have been developed and employed in the following applications: (1) real-time and adaptive pattern recognition and tracking of moving target; (2) Real-time holographic monitoring and 3D analysis of flow or temperature fields; (3) Optical testing by transversal and longitudinal wavefront shifting interference. Both the theoretical analyses and the preliminary experimented results are presented.

A realistic model has been developed for a barium titanate triple-color phase oscillator based on the mechanism of polarizabililty and quantum mechanics. It helps to explain some of the difficult phenomena of the phase oscillator. As a result, with the clear understanding, we can seek betterment of the oscillator as a photonic switch as well as a one color writing and another color displaying no cross talk advance information exchanger.

The contradictory requirements are presented to acousto- optic tunable filters (AOF) of spectral image analysis. On the one hand AOF should have high speed. On the other hand it should have good spectral resolution and wide angular aperture. Thus when AOF is fastly tuned with chirp transients, the diffracted wave intensity at different moments of transient process can considerably diverge form its quasistatic level. It means that spread function (SF) depends on the velocity of frequency tuning, i.e., it is described by 2D function with variables - wave length and velocity of frequency tuning. In Cartesian frame this dependence is presented by some surface being dynamic SF (DSF). It characterizes speed and selectivity properties of AOF. In this work DCF mathematical model was constructed and basic properties of spectral image analysis AOF were investigated. It has been established that the greatest distortions of DSF occur if velocity of frequency tuning has exceeded some critical value connected with acousto-optic interaction geometry and aperture sizes of beams. In this case the side lobes of SF will make 'false' maxima which begin to prevail over the basic. In addition under the conditions of phase mismatch DSF loses the symmetry to position of the main maximum. These effects reduce the accuracy of spectral measurements when tuning velocity is high.

We propose a complete system of blind image restoration. We have no restrictions about the blurring filter causing degradation to the original image. The blurring filter estimation part of this system consist of two main steps: the mapping of the blurred image into the Radon transform domain, and the blind blur identification in this domain. The mapping simplifies the computational complexity associated with our system from 2D to 1D domain. The blind blur identification is based on a modified optimization method which uses cumulants in the estimation of the filter coefficients. The deconvolution step is based on a least squares optimization method. Singular value decomposition technique is used in improving the optimization process in the estimation and deconvolution steps. Finally, the inverse Radon transform is computed to get the estimated restored image.

We evaluated the performance of circular optical trade-off SDF (OTSDF) and distance classifier correlation filters (DCCFs) as rotation-invariant correlation filters. Because the filters are designed using different parameters we compared the filter's performance in terms of an equivalent effect of probability of error. The use of OTSDF and DCCF filters as circular filters allows their calculation to be greatly simplified when compared to using rotated views of an object to create filters. We found that both types of filters can be used for rotation-invariant object recognition in a noisy environment. In addition, the filters generated were real-valued so they may be implemented on a variety of spatial light modulators.

In this paper a novel method of realizing digital filters in an incoherent correlator is proposed. A gray-scale complementary encoding method is used to express the positive and negative numbers simultaneously so that the digital filtering will be simplified into a convolution of the encoding images and corresponding filter mask followed by a subtraction operation, which is easily done by optical method or computer. Based on the new method a compact incoherent optical digital filtering unit (IODFU) is built up. In the IODFU a SHARP QA-1200 8.4 inch active matrix TFT liquid crystal display panel is used for displaying the gray-scale images to be analyzed and the filter masks at the same time. The images and filters can be changed by computer-controlling for different destinations. The IODFU is very compact and the processing speed can get to 12 frames per second. At the end of this paper the experimental results are given.

Based on the known diffraction property of an elementary volume grating and on the linear superposition assumption of the diffracted wave field, we propose in this paper a simple method to analyze the cross-talk noise in an angle multiplexing holographic memory. The variation of the angular selectivity of stored volume holograms due to the change of the incident angles of both the signal and the reference beams is considered. It is found that the cross- talk noise varies with the order of the retrieved hologram in an oscillation fashion. The highest cross-talk noise given in this paper is larger than that given previously and should be approximated by a piecewise linear or quadratic function of the storage capacity, rather than by a simple linear function. The effect of increasing the angular separation between the reference beams of adjacent holograms on the cross-talk performance is discussed.

The theoretical model of nonstationary diffraction on slant holographic gratings in photopolymer film is represented. The model is based on the analytical solutions of the interconnected equations of photopolymerization kinetics, diffusion of photopolymer composition components and light diffraction and describes the dynamical process of holograms reconstruction and its optical amplification. The peculiarities of the dynamics of amplitude-phase distributions and selective properties of slant gratings are discussed on the base of the numerical modeling results.

The theoretical investigation of the holographic grating pulse recording in the absorbing photopolymer materials is presented. The analytical models, describing the space-time transformation of holographic grating at the recording and the post-expositional amplification, have been obtained with taking into account the optical absorption and the diffusion processes. It is shown that the optical beams attenuation in the recording process leads to the spatial nonuniformity of index grating profile and its growth kinetics.

Polydiacetylenes (PDAs) are attractive materials for both electronic and photonic applications because of their highly conjugated electronic structures. They have been investigated for applications as both 1D conductors and nonlinear optical (NLO) materials. One of the chief limitations to the use of PDAs has been the inability to readily process them into useful forms such as films and fibers. In our laboratory we have developed a novel process for obtaining amorphous films of a PDA derived from 2- methyl-4-nitroaniline using photodeposition with UV light from monomer solutions onto transparent substrates. Photodeposition from solution provides a simple technique for obtaining PDA films in any desired pattern with good optical quality. This technique has been used to produce PDA films that show potential for optical applications such as holographic memory storage and optical limiting, as well as third-order NLO applications such as all-optical refractive index modulation, phase modulation and switching. Additionally, copolymerization of diacetylenes with other monomers such as methacrylates provides a means to obtain materials with good processibility. Such copolymers can be spin cast to form films, or drawn by either melt or solution extrusion into fibers. These films or fibers can then be irradiated with UV to photopolymerize the diacetylene units to form a highly stable cross-linked PDA-copolymer network. If such films are electrically poled while being irradiated, they can achieve the asymmetry necessary for second-order NLO applications such as electro-optic switching. On Earth, formation of PDAs by the above mentioned techniques suffers from defects and inhomogeneities caused by convective flows that can arise during processing. By studying the formation of these materials in the reduced-convection, diffusion- controlled environment of space we hope to better understand the factors that affect their processing, and thereby, their nature and properties. Ultimately it may even be feasibly to conduct space processing of PDAs for technological applications.