Optical computing is one of several emerging technologies in which DuPont is risking investment in the effort to establish a leadership position in the electronic imaging business. The decision to pursue optical computing resulted from an earlier decision to establish a parallel computing expertise as the chief building block of this new business. This paper describes the initial plans for the optics related aspects of DuPont's parallel processing venture.

An optical implementation is considered for a novel system, recently invented by the author, for accurately locating objects relatively close to an array from distributed backscatter. Processing is partitioned into handling scatterers in annuli rather than in pie shaped regions as in conventional systems which use beamforming followed by matched filtering. The advantages of the new design are discussed. Precomputing a matrix containing geometric and transmission pulse information leaves only a matrix-vector computation to be performed in real-time. Experimental results are shown for a 100 by 100 matrix-vector multiplication with a new device, a spatial light rebroadcaster (SLR). A method of constructing an optical sonar or radar is proposed using the new concept and the optical matrix-vector multiplier.

The use of photorefractive holograms in conjunction with a spatial light modulator (SLM) to realize a reconfigurable optical interconnection with very high energy efficiency is briefly discussed. In this approach, the SLM is used as a programmable binary matrix mask to encode the interconnection pattern to a coherent laser beam whereas the photorefractive crystal is used as a dynamic holographic medium to store the pattern and to efficiently diffract the readout beam into the selected channels at high speed. We report recent experimental results on issues such as energy efficiency, reconfiguration time, contrast, and uniformity.

The implementation of fast Fourier transforms (FFT) via the use of residue factored look-up tables (FLUT) is investigated. The principles of FLUTs are reviewed, and a gate-level pipelined adder architecture is presented. The basics of the quadratic residue system (QRNS) are then discussed, and FLUT-based gate-level pipelined architectures are presented for binary-to-QRNS and QRNS-to-binary converters, as well as for FFT butterflies. The Despain small integer approximations are used to represent the FFT complex rotations each of which is expressed via a linear combination of common angles. The QRNS FLUT FFT performance is measured via a normalized mean square error (MSE) figure which is estimated via computer simulations performed for 16-, and 32-point QRNS FFTs in conjunction with various input signals and different approximation accuracy. Based on these results the largest FLUT FFT order, for which an MSE of <10^-8 can be supported, is estimated. The overall system gate complexity is then calculated and compared with that required by the equivalent conventional digital implementation.

For many applications, such as signal and image processing, computer vision, and artificial intelligence, the current achievable performance is much lower than that needed. Von Neumann models cannot achieve computational rates equivalent to billions of operations per second that will be required for these applications. Therefore, parallel models of computation must be used. One attractive concept for future computer architecture is the data-flow model of computation. Theoretically, maximal concurrency can be exploited using such a model. However, the lack of efficient implementation has prevented its wide use. This paper examines how optical systems can be used to break the bottlenecks that conventional electronic implementation imposes and presents a high-performance and scalable optical data-flow architecture. The architecture exploits the high degree of connectivity and inherent parallelism in optics for implementing a highly parallel instruction-level data-flow multiprocessing system. Both architectural and implementation issues are considered.

We have observed waveguiding in thin films of polymer gelatin on GaAs, LiNb0₃, glass and aluminum substrates. A graded index profile can be induced in the gelatin layer and tuned by wet processing. This makes it possible to form waveguides on any smooth surface. Locally sensitizing the gelatin waveguide with ammonium dichromate allows us to integrate single and multiplexed gratings on the same substrate to perform various functions for optical interconnects and signal processing. A waveguide grating coupler that converts free space TEM₀₀ laser light to a two dimensional spherical guided wave with 50° angle of divergence has also demonstrated. A passive broadcasting network can be formed using this new technology. Further plausible applications such as WD(D)M local area network, optical interconnection, and optical computing are also presented.

Concurrent computing systems are effective only if the data rates to and from processing elements are sufficient to keep the processing elements occupied. The inherent parallelism of optics makes it a natural candidate for information transfer while the existing technology of silicon very large scale integrated (VLSI) circuits is the most suitable for performing computational tasks. In this paper we present the use of fixed optical interconnects for information transfer in electronic based concurrent computing systems. This capability is based on the development of electrically addressable spatial light modulators (SLM's) fabricated as part of integrated circuit devices.

Novel electron trapping (ET) materials capable of performing optical parallel Boolean logic operations are described. An application to binary full addition based on a newly developed parallel algorithm is discussed, and experimental results are presented.

The increasing need of computing power has generated a renewed interest in truth-table look-up processing. But as the number of input digits increases the size of the resulting truth-table increases so fast that the required number of reference patterns may become unmanageable. Usually residue number system is used to solve this problem. However residue arithmetic processors suffer from several disadvantages. The most significant one is the time delay involved in encoding the inputs to residue representation and decoding the final result to a binary or decimal representation. In this correspondence', we proposed a simple way to reduce the truth-table for addition and multiplication. Instead of producing sum's truth-table, a carry's truth-table is produced. Carries in each bit can be get simultaneously by truth-table look-up technique. To get the final result, a half-adder is required. Each bit of the result is the half-adder of three inputs, one is the carry, the other two are augend and addend respectively. In multiplication, the product is expressed as a composition of addition and square, the result truth-table of square are much smaller than the table directly constructed from product, so the number of reference patterns required is greatly reduced. Comparisons are given between two kinds of tables, directly constructed from addition and multiplication or produced in the way described above. In the experiment, PROM is used as a logical device to implement logical based pattern recognition. Experimental results are also demonstrated.

Two novel architectures for implementing the LMS algorithm optically are introduced. Convergence of the LMS algorithm is analyzed and the effect of the analog inaccuracies are considered.

A method is described for optimizing the power coupling efficiency between a single fiber or bundle of fibers and either an optical component (such as a detector, coupler, splitter, etc.) or another fiber or fiber bundle. The technique described is to either (i) match the cross sectional areas of the components to be butt-coupled or (ii) decrease the diameter of the active area of the fiber or fiber bundle to be lens-coupled so that the image size and numerical aperture of the lens are optimized. In either case the desired result is accomplished by controlled etching of the ends of each of the appropriate fibers involved in the coupling. We have shown that the loss in optical power associated with pre-etching a fiber before it is epoxied in a resilient ferrule connector is less than 0.5 dB even when the fiber cladding is completely removed in the etching process. Thus, efficient coupling has been demonstrated, for example, between a bundle of seven etched fibers and a single unetched fiber (i.e., a fan-in coupler). Although the technique is primarily of wider application with multi-mode fiber, it has been shown to work equally well for single-mode fiber.

The Bimodal Optical Computer algorithm for solving systems of linear equations with bipolar numbers is presented along with the experimental results. In addition, the experimental results of solving for the inverse of a matrix is reported.

Artificial intelligence (AI) concepts have been used to build a program capable of following and marking roads, highways, and waterways in digital imagery. Such a program is capable of ignoring unimportant objects such as lamp post shadows or vehicles on a road, or ships and bridges on a river. The concepts of experience, learning, decision making, and sensing of the upcoming "environment" have been incorporated into the code.

We examine the effect of varying ground clutter on the performance of optical correlators. Comparisons are made between the classical matched (CMF), phase-only (POF), and binary phase-only (BPOF) filter. A new definition of the signal-to-clutter ratio (SCR) in the input scene and the signal-to-noise ratio (SNR) in the correlation plane is presented. Simulations show the POF and BPOF always outperform the CMF. The BPOF performs best in scenes with low SCR ratios (high clutter) using the new metric. The POF performs best for all levels of SCR when a block is placed in the filter plane.

Two algorithms are compared for the design of Phase-Only Synthetic Discriminant Functions. The first, known as the Successive Forcing Algorithm (SPA), uses a Gerchberg-Saxton type of iteration. The second, known as the Relaxation Algorithm (RA), was proposed recently by Jared and Ennis. Preliminary simulation results are included to facilitate a comparison between the two.

A joint transform correlator is demonstrated that utilizes a high-speed optically addressed spatial light modulator in the Fourier plane consisting of an hydrogenated amorphous silicon photodiode and a ferroelectric liquid crystal modulator. Results are shown for binary-amplitude modulation using a smectic C* ferroelectric liquid crystal operating at 4600frame/s, 20:1 contrast ratio, and 38lp/mm.

We compare the performance of an optical binary phase-only correlator using several nonlinear filters for preprocessing. An image of an airfield with different but similar appearing planes is used as the input data. The performance of the correlator was measured by examining several parameters in the correlation plane. Performance criteria are the signal-to-clutter ratio, the signal-to-noise ratio for objects within the same class, and the variance of the correlation peak within a class. Input images to the correlator were preprocessed using nonlinear filters based on both median filtering and adaptive filtering. The preprocessing improved the performance of the correlator by supressing noise while preserving the edges of the input image. The SNR changes with the position of an object due to diffraction from the pixels of a spatial light modulator.

There has been a growing interest recently in the use of Phase-Only Filters (POFs) and Binary Phase-Only Filters (BPOFs) for optical pattern recognition. While BPOFs can be obtained by binarizing the phase of the classical matched filters, one wonders if the filter performance can be improved by allowing more phase levels. This paper provides a theoretical analysis of the increase in output Signal-to-Noise Ratio (SNR) as the number of phase levels is increased.

Ternary phase-amplitude filters (TPAF, encoding the modulation states -1, 0, and 1) are being developed for correlation both in theory and in practice. They are attractive due to their few discrete modulation levels which facilitate efficient electronic storage for on-line systems and allow implementation in real time with available spatial light modulators (SLMs), as recently demonstrated using magneto-optic devices. Simulations have demonstrated that effective smart TPAFs can be formulated to address class discrimination goals. We report experimental demonstration of significant increases in correlation discrimination achieved with TPAFs designed with the transform ratio technique. Results are in substantial agreement with theory based on computer simulations, and verify the practical implementation of improved discrete-level filters using magneto-optic SLMs.

The operational characteristics of deformable mirror device (DMD) spatial light modulators for image processing applications are presented. The two DMD pixel structures of primary interest are the torsion hinged pixel for amplitude modulation and the flexure hinged or piston element pixel for phase modulation. The optical response characteristics of these structures are described. Experimental results detailing the performance of the pixel structures and addressing architectures are presented and are compared with the analytical results. Special emphasis is placed on the specification, from the experimental data, of the basic device performance parameters of the different modulator types. These parameters include modulation range (contrast ratio and phase modulation depth), individual pixel response time, and full array address time. The performance characteristics are listed for comparison with those of other light modulators (LCLV, LCTV, and MOSLM) for applications in the input plane and Fourier plane of a conventional coherent optical image processing system. The strengths and weaknesses of the existing DMD modulators are assessed and the potential for performance improvements is outlined.

The effect of variation of design parameters on the performance of phase-only and binary phase-only correlation filters is presented. Correlation performance is analyzed from the perspective of optimized extraction of a target from an input scene consisting of the target, a natural background, and additive noise. It was found empirically that the dominant factor affecting the filter performance was the target image background level and that the optimum choice depends on the relative values of the target average value and the background average value in the input scene. The behavior is explained analytically by describing the target object in terms of a target silhouette function that specifies the target region, and a target detail function describing the target variations within the silhouette. The optimum choice for the target image background is shown to be that which results in the addition (rather that subtraction) of the detail and silhouette autocorrelations.

A rotation and scale-invariant optical correlator has been developed, assembled, and evaluated for use as a multiple target recognition system. The system consists of an optical correlator with magneto-optic spatial light modulators (MOSLMs) at the input and filter planes and a vidicon at the correlation plane. A COMPAQ 386 (IBM compatible) personal computer with a frame grabber board is used to acquire, binarize and load binary amplitude-only video images to the input MOSLM, to write sequential stored binary phase-only filters to the filter MOSLM, and to sample and statistically analyze correlation plane data in order to locate and recognize objects of interest in the input scene. The sequential correlations and output data samples are obtained at near video rates (15 per second), allowing multiple targets at any scale and in-plane rotation in a given input image to be located and classified in seconds. Experimental results indicate that this system can correctly identify objects within its target class more than 90% of the time, even in the presence of severe clutter and noise.

We present a parallel optical method of multiplying a two-dimensional input array by an arbitrary fourth-rank tensor to produce a 2-D output array. This process is equivalent to a crossbar interconnection of the input and output arrays. The algorithm works by arranging the interconnection tensor into a 2-D phase-coded image. The input is pre-coded by transmission through a conjugate phase mask, and then optically correlated with the connection image. The correlation plane contains the output array values in a noisy background. We show how the output signal to noise ratio (SNR) can be traded in for connection image size and complexity, and give theoretical calculations of average output SNR, along with computer simulations supporting these predictions. We describe a proposed reconfigurable optical interconnection system using this matrix-tensor multiplier implemented with four-wave mixing in photorefractive materials.

A binary phase hologram for optical interconnects is divided into two blocks and binary phases of each block are decided by a simulated annealing algorithm. Such a binary phase hologram showed high diffracion efficiency and was able to produce asymmetrical diffraction orders. Performance limitations of these holograms are studied.

A magneto-optic spatial light modulator (MOSLM) has been proposed for use as a Fourier plane filter in a coherent optical correlator. Its binary nature and limited, presently small, space-bandwidth product constrain any filter design. Although binary quantization allows a maximum number of Fourier values to be coded, quantization and reconstruction error is high except in a few cases. To reduce these errors, a cell-oriented binary coding technique, the delayed-sample method, is used. Three cell sizes are considered: 2 x 1 pixels, 3 x 1 pixels, and 4 x 1 pixels. Through coding, a 2 x 1 cell can realize three real values {-1,0,1} as opposed to only two {4,1} for binary quantization; however, there is a trade-off in the number of Fourier values that can be coded. For a 2 x 1 cell the number is reduced by one-half. A 3 x 1 cell can realize seven complex values, but with a one-third reduction in the number of coded Fourier values. Finally, a 4 x 1 cell is capable of realizing nine complex values with a one-fourth reduction in the number of coded values. The trade-off between quantization error and number of Fourier values coded is examined qualitatively using a 128 x 128 MOSLM. Reconstructions from coding using different cell sizes are compared to reconstructions from binary quantization. In addition to coding, hologram replication is used to improve reconstruction error. Sampling issues relating to the size of the filter response are also discussed.

This paper presents a description of the Litton/Semetex magneto-optic spatial light modulator (MOSLM) switching characteristics and a drive design approach to achieve high frame rates. This drive approach minimizes MOSLM write power and provides higher operating current margins than standard drive approaches at all frame rates. Test results are presented for driving a 128 by 128 MOSLM at 2 Khz frame rate with 50% data patterns and no pixel errors.

A nonlinear matched filter based optical correlation is presented. The nonlinear filter is produced by applying a nonlinear transformation to the linear matched filter.

Fourier-Mellin filters are used to generate invariant feature descriptors. The position of the maximum correlation output for each filter is used to create a distance vector. This vector is invariant under translation, rotation, change of scale and intensity of the input object. This method is applied to seven objects and the resulting, vectors are fed to a neural network for recognition. Of the seven objects, five are used to train the network. A three-layer feed-forward network, trained with the back-propagation algorithm is employed. Results show that distance vectors are suitable inputs for recognition by a neural network. The network learns the associations and recognizes the objects.

The maximum mean square projection (MMSP) filter is proposed for distortion invariant recognition. It classifies images according to the energy projected onto an N-dimensional subspace. The MMSP filter is implemented in an updatable optical correlator where correlations from multiple filters are time integrated on the output detector array.

The synthetic estimation filter (SEF) combines a set of input images into a much smaller set of filters. Those input images correspond to views of a known object taken with a variety of values of the pose parameter to be estimated. The response of the filter to variations in the pose parameter is tailored so that estimation can be done to a precision considerably finer than the intervals between filters in the pose space. We here extend the technique to the joint transform correlator; as a reference we use the composite image from which the SEF is conventionally calculated. Precautions are necessary for practical reasons. For example, the SEF algorithm may call for negative values in the image, an unrealizable condition with amplitude encoding, whereas with phase-mostly filtering in the SEF, negative image values are easily accommodated.

Optical temporal pyramidal image processing based on liquid-crystal televisions (LCTVs) and noncoherent illumination is proposed. compact non-coherent optical tracking novelty filter and its use as a temporal pyramidal filter are demonstrated. Some preliminary experimental results are presented.

An important subset of pattern recognition applications permit the representation of data by characters which have been optimized for the type of data and the type of search to be performed. An example of this is the search for biologically important patterns within the sequences of nucleotide subunits of deoxyribonucleic acid (DNA). In this case the four distinct subunits of DNA must be represented and "wildcard" or metacharacters are needed to permit flexible searches for sequence patterns. Due to the rapidly increasing availability of DNA sequence information, more rapid and interactive analytical techniques are needed to make full use of this data. This study seeks to design optimal characters for use in an optical correlator recognition device. Characters which are compact, easily distinguishable and compatible with current coherent light modulators have been designed. Preliminary work on these representations has been guided by computer modeling of the optical recognition process. Promising characters have been tested experimentally in a VanderLugt system. The use of laser printers and photo-typesetters to prepare original test images will be discussed.

Vander Lugt correlators are limited by the sensitivity of their Matched Spatial Filters (MSFs) to image scale, object orientation, and aspect angle. For real-world applications, the number of filters required to identify and track an object, in any relative position, is large. A solution to this problem is the use of an addressable filter (i.e. a Spatial Light Modulator) which preserves the system's optical alignment even though the filter has changed, and feedback from the correlation plane which servos the input scene display. Perkin Elmer has built a correlation system capable of tracking a real target, in any rotational orientation, over a 16:1 change in scale, in real-time. The system uses two Liquid Crystal Light Valves (LCLVs), optically addressed by CRTs at video frame rates. One serves as the correlator's input and operates in an amplitude mode; the second functions in a phase-only mode and serves as the Matched Spatial Filter. Three spatially multiplexed Binary Phase-Only Filters (BPOFs) operate on the fourier transform of the scene in parallel. Each filter acts independently, and produces a correlation spot. The center BPOF is computed from a slightly magnified image of the target object, and the adjoining filters are based upon correctly scaled, but slighty counter-rotated images of the target. By cross-referencing the three correlation outputs, the type of scene distortion (scale and/or rotation) can be determined, and the input scene CRT raster can be adjusted to maintain the optical match required for continuous correlation. This allows a MSF to operate over a much wider range of conditions than previously possible. By monitoring the scale of the scene CRT's raster, the system can determine when the next addressable filter must be called up from memory and displayed to continue correlation. Two computers run the software for the system, which contains image processing programs to capture input video images, generate filters, rotate and scale the scene CRT's raster scan to preserve correlation performance, change filters when the limits of raster scan size are exceeded, and track the target. The system works in conjunction with a remote gimballed telescope platform that provides the input video imagery. Tracking error information from the computers is telemetered to the platform to maintain the line of sight to the target.

This paper provides an overview of architectures for nonlinear transformations used in optical image processing. The emphasis is on the decomposability of certain nonlinear transformations as serial (cascaded) and parallel (multiplexed) combinations of elementary operations.

A new method is proposed to detect and enhance such features as object bound-aries or line segments in a noisy gray-scale image. This method utilizes directional information at each point in the input image. The input image is convolved with a 2-D kernel, discussed below, which is rotated through 360 degrees, either continuously or discretely in a fairly large number of steps. As the kernel rotates, the convolution output is measured and the maximum, minimum, and mean values at each point (as a function of rotation angle) are stored in a computer. Once these values are obtained, a class of image processing operations can be performed. In an optical implementation of the processing operation, it is necessary to physically rotate a mask in the optical system. However, this is much faster than effecting an equivalent kernel-rotation operation with a digital image processor.

NASA's planning for the future exploration of the solar system includes missions requiring extremely powerful information processing systems. Optical information processing can play a key role in meeting these needs due to its high throughput and fault tolerance, low power consumption and weight. Several major optical information processing research projects such as optical pattern recognition, neural network, AOTF imaging spectrometer, spatial light modulator, and optical matrix processor, are under development at JPL, AMES, and JSC. These optical processing research efforts and their applications in NASA's various planetary exploration missions are discussed.

The Kittler-Young (K-Y) transform is a nonparametric method for feature extraction. The important property of the K-Y transform is that the information of class variances and mean squares are utilized optimally in feature selection. A joint transform correlator (JTC) is used to extract the features of the K-Y transform from input images optically. Making use of these features, classifications are performed on a microcomputer.

Although joint Fourier transform processor is effective for the application to correlation operation, it can also be used as a generalized coherent image processor. In this paper we shall show that blurred photographic images due to linear motion can be restored with a joint transform processor. The major advantage of using the joint transform processor (JTP) for the image deblurring must be the avoidance of spatial filter synthesis. In other words, the deconvolution function for the image deblurring can be directly implemented at the input plane of a JTP, and the difficulty caused by the singularities of an inverse filter can be avoided. To generate a bipolar deconvolution function, we have encoded two sets of π phase shifted sinusoidal grating of the same spatial frequency. Computer simulation for restoration of linear-motion-blurred images are given to test the feasibility of the scheme.

An image enhancement method-the real-time image cosine transformation is presented. The photo intensity signal from an X-rayogram which corresponds to a single period of the cosine function is coded through a cosine transformation processor(CTP)1, and then decoded and made reappearance of the Image by means of a television mornitoring system. The image coding and the decoding are discussed. The step wave processing by the CTP 18 shown, and the experimental results of X-rayograms processing in the early stage of pathological changes are given.

For the quarter of a century I have been involved in optical computing we have put forward a rather consistent set of arguments and beliefs which have led to a great increase in our numbers, knowledge, funding, and capabilities. I believe we must now challenge these basic tenants of our faith in the name of honesty and in the hope of a brighter future.

Optical symbolic substitution offers general purpose optical processors for numeric, logic, and image processing functions. Its realization on an optical correlator is most attractive. This paper addresses the optical laboratory realization of the symbolic substitution steps of recognition and substitution with attention to the effects of the beam balance ratio (K-ratio) used when synthesizing the matched spatial filters for a symbolic substitution correlator. Specifically, low spatial frequencies should be emphasized in the substitution step and high spatial frequencies should be emphasized in the recognition step.

Two types of optical associative memory for two-dimensional image retrieval using nonlinear correlation is described. The product between the input image and the stored images is obtained by nonlinear correlation technique which has a superior performance compared with the conventional optical correlation techniques in the areas of the light efficiency, the correlation peak to sidelobe ratio and correlation spot size. We show that for some noisy inputs, it is necessary to use nonlinearity at the Fourier plane to improve the correlation SNR. Using this technique, better quality images can be reconstructed and the need for optical gain and the optical feedback may be eliminated.

We describe and present results of an optoelectronic neural network processing system. The system uses an algorithm based on the Hebbian learning rule to memorise a set of associated vector pairs. Recall occurs by the processing of the input vector with these stored associations in an incoherent optical vector multiplier using optical polarisation rotating liquid crystal spatial light modulators to store the vectors and an optical polarisation shadow casting technique to perform multiplications. Results are detected on a photodiode array and thresholded electronically by a controlling microcomputer. The processor is shown to work in autoassociative and heteroassociative modes with up to 10 stored memory vectors of length 64 (equivalent to 64 neurons) and a cycle time of 50ms. We discuss the limiting factors at work in this system, how they affect its scalability and the general applicability of its principles to other systems.

A new bidirectional optical associative processor is described for searching a hierarchical database that is stored as an adjacency matrix. The paper discusses how the processor can answer relatively complex queries on a knowledge base when the queries are formulated as combinations of set closures, unions, intersections, and complementations. Thus, a processor that performs general set operations results, as well as a system that can answer various knowledge base queries and guide a knowledge base search. These are new operations for associative processors that increase their utility. This new associative processor operates on entities and their attributes. It can be viewed as a type lattice processor (since the entities and attributes form a hierarchy known as a type lattice), as a closure processor (since it performs closure operations that list all attributes of an entity [or entities] or all entities with a given attribute [or attributes]), or as an adjacency processor (since the connection matrix used stores adjacent associations of attributes and entities).

There has been a resurgence of interest in artificial neural networks. However, dedicated hardware implementation is still a bottleneck for the development of significant applications. The neurons and the synapses are the two key components in a neural network. Synapses are used to provide massive interconnections between different neurons. The computational capability of a neural network depends on its storage capacity and speed. The storage capacity is related to the number of synapses (interconnections) between individual neurons. The information is stored in the strength and the distributed pattern of the interconnections. The 2-D nature of optics with inherent capability to provide large connectivity is particularly attractive for implementing neural network models.

The difference-squared error algorithm provides an effective means for discriminating between objects of interest and structured background features. This paper briefly discusses the advantages of this algorithm over conventional correlation based recognition techniques. The output of a real-time acousto-optic processor which computes the difference-squared error between an input data stream and a reference function is presented. The output of this acousto-optic implementation is then compared to the desired theoretical results. Finally, future applications of this architecture for real-time two-dimensional pattern recognition is discussed.

An important difference between most Acousto-optical (AO) and imaging systems is coherent vs. incoherent light. The Transfer Functions in the presence of the primary aberrations underscore these differences. Two AO types, the Acousto-Optical Tunable Filter (AOTF), and the Acousto-Optical Dispersive Light Filter (AODLF), accept incoherent light. The AOTF is an electronically programmable monochromator. The AODLF is an agile spectrometer which, can be used with wide angle optics to comprise an unusual optical system.

The proton exchange technique has been used to realise monomode waveguide lenses and Bragg cells in Y-cut lithium niobate. These results are discussed, and the performance of a fully packaged integrated acousto-optic correlator fabricated using the above components is presented. Recent experiments to further optimise, device performance are described.

Acousto-optics deals with the interaction of sound and light. When an acoustic wave propagates in an optically transparent substrate, it produces a periodic modulation of the index of refractions through the elasto-optical effect. This creates a moving phase grating which may diffract portions of input laser light into one or more directions, and is known as acousto-optic diffraction. This phenomena is a base for various devices, such as: deflectors, modulators, frequency shifters, filters, etc.

A novel concept for implementing a high resolution spatial light modulator using a thin slab of photorefractive crystal is described. Experimental demonstration as well as an analysis of the operation and performance of the device are given.

The liquid crystal optically addressed spatial light modulator (OASLM) is analyzed from a sensitivity viewpoint. Optimum material properties are discussed. The effect of different operating waveforms on the OASLM sensitivity is examined. Some experimental results on the sensitivity limitations of the GaAs/liquid crystal OASLM are reported.

Novel multiple quantum well (MQW) optical light modulator structures using the quantum-confined Stark effect induced by a propagating surface acoustic wave, are described. The proposed structures are shown to generate electric field strengths in the range required to produce significant shifts in electroabsorption and index of refraction for the AlGaAs-GaAs system. The operation of a Bragg modulator/switch, which utilizes the QCSE to enhance the diffraction effects and forms an acoustic phase grating produced by the SAW beam, is also discussed. The schematics of an integrated spatial light modulator based on enhanced Bragg modulation is presented. In addition , the application of spatial light modulators and the optical devices in the realization of an optical correlator is discussed.

In an optically addressed SLM there are many possible trade offs in device performance. Typically performance is altered by modification of the detailed device design, and also by modification of the electrical drive scheme. In this paper we show how it is also possible to change the device performance by changing the polarisation characteristics of the incident read light. We discuss the operation of an MRC amorphous silicon/nematic SLM under two distinct polarisation conditions, evaluate the device performance and relate this to potential applications.

A new type of spatial light modulator, based on the surface plasmon resonance effect, is proposed and demonstrated. The theory and design of these devices is discussed, as are their special features and performance potential in coherent optical systems. The prototype devices use liquid crystal as the active material, and encouraging results have been obtained for speed (2 ms), contrast (100:1), and resolution (20 μm). Higher performance devices using other structures are proposed; these include multi-quantum-well diodes, Langmuir-Blodgett films, and reversible electroplating cells.

In this paper we present enhanced four wave mixing with a sinusoidal phase shift introduced in the forward propagation pump beam. The intensity distribution on the crystal is consequently modulated by a sum of frequencies each scaled and phased according to an ordinary Bessel function. The frequency analysis of the photorefractive band transport model shows that the imposed sinusoidal phase shift provides information on both phase and amplitude of the space charge field.

The exponential gain of two-beam coupling in InP:Fe crystals under dc fields is a function of the pump beam intensity at a given temperature. We propose two methods in order to extend the sample volume working under optimum conditions. The first method is based on a negative thermal gradient and the second uses an auxiliary incoherent beam. Theoretical simulations predict an integrated gain of 15cm^-1 and 10cm^-1 respectively (for λ=1.06pm, λ=5μm and 0_° =10kV/cm) in thick crystals (L=5mm) . Preliminary experimental results confirm that high gain performances can be achieved by using these methods.

A computer interfaced, 5000 X 7500 picture element, 20,000 X 30,000 addressable position, spatial light modulator (SLM) system for ultra high resolution optical processing applications is described. The system consists of a host computer such as an IBM PC, electronics and optics with beam position feedback to rapidly and precisely scan a focussed laser beam over the writing area of the SLM, and an erasable, editable, SLM with nonvolatile storage of the image written by the scan system. Because the image is stored by the SLM, no periodic electronic refresh is required. The resolution and information content achieved by this SLM system is higher than that of professional quality color film. Potential applications include optical correlation, optical processing of photographic quality information, electronic holography and optical computing.

Described here is a previously unreported optical characteristic of magneto-optic film which is in the "heterogeneous", or "third" stable magnetization state. In this largely disregarded state, such material displays a serpentine-like pattern of oppositely magnetized domains, whose individual size is on the order of ten microns in width. The optical effect is seen most clearly when a single large unpatterned magneto-optic device is struck at normal incidence by a linearly polarized laser beam. The magnetooptic film, when in this third state, splits the incident beam into an undeviated central beam whose angle and quality of polarization matches that of the input beam, as well as first and higher order diverging rings, whose angles of polarization lie at 90 degrees with respect to the input (and central output) beam's polarization. Experimental results are presented which detail the optical output. A mechanism for the effect is described, and applications of this "third" state output to binary Magneto-Optic Spatial Light Modulators are discussed.

Calculations based on theoretical models of metal-insulator- semiconductor structures show that moderate applied potentials may cause sufficiently large induced inversion charge carrier densities at the semiconductor surface to yield reflectivities approaching 100 percent at the insulator-semiconductor interface. Using p-type silicon as the semiconductor material, positive gate potentials up to 10 volts applied to the metal predict reflectivities from approximately 15 percent to nearly 100 percent. The dependence of doping concentration, insulator thickness and gate voltage are shown.

The development of a reflection mode nematic field effect a � Si:H LCLV is described. The amor-phous silicon photoconductor layer is deposited by glow discharge without doping. The configuration al-so consists of dielectric mirror as CdS LCLVs do. The bias frequency is in the range of 500Hz to 10KHz. MTF, sensitivity and time response of the device are measured. High resolution of > 501p/mm and switching time less than 25ms are achieved. Also reported is an equivalent circuit mode which has been compared with experimental results.

A submicron metal grid mirror was incorporated into the structure of a silicon liquid light valve. In our experiment, a 0.5 μm-period, aluminum wire grid mirror was used in conjunction with a 90°-twisted nematic configuration. A sharp threshold with a peak-to-threshold ratio of 3:1 was experimentally observed. The threshold intensity level was tunable from 2.5 μW/cm² to 50 μW/cm² by changing the bias voltage from about 5 to 25 volts. These properties allow the polarization-sensitive mirror-based LCLV to be used as a non-linear SLM with applications in optical computing (bistable device) and adaptive image thresholding. The device can also be used as an interfacing device for an optical co-processor as well as for optical implementation of Hopfield-Anderson association and phase conjugation.