A need for very high probability of detection in modern passive collection systems has stimulated development of ever-wider receiver bandwidths. As the receiver bandwidths increase, some portions of the RF spectrum will begin to have a probability of signal overlap approaching unity. Wideband receivers, when operating in these environments with increasing signal threat densities and new modern modulation techniques, must incorporate computationally intensive algorithms within the receiver architecture to ensure reliable performance. Acousto-optic processing subsystems offer a real-time solution to several of the computationally intensive signal processing functions required in wideband passive collection systems. Two such acousto-optic subsystems for channelization and direction finding will be discussed in this paper. The development is being sponsored by DARPA through the Naval Research Laboratory.

This paper describes an acousto-optic (AO) processor that offers a small, lightweight solution to detecting and analyzing wide-bandwidth, spread-spectrum signals. The processor is being developed for insertion into an existing electronic support measure (ESM) test-bed. The correlator will have a processing bandwidth of 500 MHz and will be used to detect direct- sequence phase modulated (PM) signals, frequency-hopped signals, chirps, and impulse signals. An in-line AO correlator is the heart of the processor and is used for detecting wideband activity. Subsequent digital processing, including Fourier transformation, will be used to determine center frequencies, bandwidths, and band shape. Theoretical operation of the correlator is discussed along with descriptions of the radio frequency (RF) interfaces and digital post-processing.

An Acousto-Optic (AO) Range-Doppler Processor is described that is designed to interface to an advanced ground-based radar system developed by the U.S. Army Missile Command (MICOM). Demonstration of this optical processing technology in the MICOM radar is the primary objective of this DARPA-sponsored Transition of Optical Processors into Systems (TOPS) program. This paper presents a description of the MICOM radar system, highlights the design of the AO Range-Doppler Processor, and describes the required radio frequency (RF) and digital electronic interfaces to achieve real-time operation in the MICOM radar. Insertion plans for this TOPS program are also summarized.

The time-and-space integrating (TSI) acousto-optic approach for real-time synthetic aperture radar (SAR) imaging was developed for applications with severe power and size constraints. Compactness and low power consumption are achieved by performing the computationally intensive operations in the analog optical domain. The required SAR imaging filters are updated rapidly through electronic programmability and downloaded to the optical system via acousto-optic Bragg cells. Under the DARPA TOPS (Transition of Optical Processing into Systems) Program, a rugged prototype of this concept is being developed for integration with high performance SAR platforms. In this paper the TSI approach is reviewed and simulation results are presented which demonstrate the concept's ability to efficiently focus high resolution SAR phase history data across a large target area, while compensating for large amounts of range migration.

The development and airborne demonstration of a compact realtime optical processor for synthetic aperture radar (SAR) image formation under the DARPA TOPS program is described. The ERIM spotlight mode SAR system and its processing requirements are presented. It is shown that a 2-D Fourier transforming time-integrating interferometrically based optical processor is an attractive solution to the processing requirements. The optical processor uses a modulated laser diode for radar signal insertion, crossed acousto-opto scanners for 2-D scanning, a modified Köster interferometer for fringe generation, fast detector arrays for light detection and integration, and accumulating frame grabbers to build up the dynamic range of the image. Analysis, simulation, and laboratory experimental results are presented.

A highly parallel correlation system concept has been developed to rapidly locate data stored as free form text on an optical disk. A simulation using realistic component noise, and media characteristics predicts correlation outputs for an optical disk based data retrieval system using text data recorded in varying formats. Projected performance of a breadboard system is included.

Acousto-optic spectrum analyzers have found wide application in signal processing as a result of their ability to provide wide bandwidth, rapid transforms with relative ease. In the past the dynamic range of acousto-optic spectrum analyzers has been limited to approximately 30 dB by the dynamic range of the CCD sensor array. Recently CCD sensors with non-linear response have been developed with dynamic range in excess of 60 dB. In order to fully utilize the dynamic range of these sensors, it is necessary to properly weight the beam profile illuminating the Bragg cell. We examine the effect of beam weighting on the resolution and dynamic range of optical spectrum analyzers. Because of its ease of implementation, particular attention is paid to the truncated Gaussian beam shape. The effects of acoustic attenuation in the Bragg cell and amplitude and phase variations on the beam profile are examined. We find that beam weightings which result in increased dynamic range do so at the expense of decreased frequency resolution.

Photonic Systems Incorporated is currently fabricating a Multichannel Acousto-Optical Spectrometer (MCAOS) for NASA Goddard Space Flight Center. This instrument will be used as a frequency channelized radiometer for radio astronomy spectroscopy. It will analyze the spectrum of four independent radio frequency (RF) channels simultaneously and has potential for eight to as many as sixteen channels. Each channel will resolve the RF spectrum to one megahertz within its 1000 megahertz band. Dynamic range exceeding 30 dB will be achieved by quantizing detector photo-charge to 12 bits and accumulating data for large periods of time. Long time integration requires an optical bench optimized for stability and the use of temperature stabilization. System drift due to speckle interference is minimized by using a novel polarization switching Bragg cell.

Hardware implementation of the steepest descent algorithm, as applied to multichannel adaptive jamming cancellation, requires the realtime correlation of wide bandwidth signals from multiple input channels. The described optical system uses a single-channel acousto-optic (AO) deflector as an input device for the adapted main antenna signal, where multiple jamming sources mask the target return, and a multichannel AO deflector as the input device for an array of auxiliary antennas, each receiving jamming energy. A time-integrating correlation between the main and auxiliary channels is calculated optically and produces an update to each weight function (stored in computer memory) in accordance with a steepest descent algorithm. The updated weight functions are optically reconstructed and used to tap a multichannel AO delay line, which carries the information from the array of auxiliary channels. A spatial sum of the output from the weighted delay line yields an estimate of the noise in the main channel. The multichannel optical time-integrating correlator has demonstrated realtime parallel computation of the correlation between two wide bandwidth auxiliary channels and the adapted main channel.

One of the most successful optical signal processing applications has been optical architectures for converting synthetic aperture radar (SAR) data into images of the radar reflectivity of the ground. Pattern recognition using optical correlator technology has also been extremely successful, and generalized multiple filter techniques have allowed the implementation of invariant target recognition systems. A simple non-linearity in the form of an optically addressed spatial light modulator to remove random phase terms has enabled the cascading of these two architectures. Experimental verification of this cascaded non-linear optical processor using SAR data from the Shuttle Imaging Radar-A mission is presented. SAR images are formed and used as the input to an optical matched-spatial-filter correlator that successfully recognized ground features.

An optoelectronic processor of synthetic aperture radar (SAR) signals is discussed. They have a high azimuth resolution, a great number of processing parallel range channels, real-time regime, and little energy consumption. The usage of the strip LED array with a received signal demultiplexing and distribution between the LED element as the signal input device is optimal in terms of the processor's simplification and energetics. It is shown that this SAR image forming processor can be constructed as a compact optoelectronic hybrid scheme without optical elements. Theoretical and experimental results of the investigations are presented.

In the last few years, a considerable effort in the optoelectronics research field has been spent for the development of a number of guided-wave active and passive components, such as laser diodes, electrooptic modulators, acoustooptic transducers, photodetectors, microlens array, and so on, for fabricating optical devices and circuits for signal processing and computing. The interest related to optical processors is particularly due to a lower power consumption, reduced size, cost and weight, and high throughput with respect to the corresponding electronic processors. In particular, synthetic aperture radar (SAR) applications are well suited for an optics-based processing technique implementation, because the synthesis of the object image, performed by correlating the received radar signal with a reference signal, is equivalent to the optical reconstruction of the Fresnel diffraction pattern of the same object, illuminated with coherent light. Guided-wave optical processors, including acousto-optic transducers and CCD cells, can be successfully applied to the reconstruction of two- dimensional images by using both spatial and time integration. In this paper, we present the theoretical investigation, design, and simulation of a new LiNbO₃ guided-wave optical correlator suitable for real-time SAR applications. It is based on a complex interferometric structure, involving four aperiodic phase-reversal traveling wave modulators. The electrode structure is designed in order to reproduce the product signal between the received and reference voltages, which is then time-integrated by a suitable photodetector. The filtered signal coming from the detector is proportional to the final correlation function, which can be electronically registered and multiplexed on a two-dimensional matrix by sum-and-shift procedure. Thus, the processor performs the correlation function between the reference signal and the received signal when they are applied to laser diode and to the electrodes as driving voltage, respectively.

We present a novel architecture for an optical guided-wave recirculating (feed-backward) delay line, which is suitable for planar implementation, and is capable of performing very high speed optical processing. The chain matrix has been demonstrated in the z-transform domain and its identity with conventional fiber-optic architecture has been demonstrated. This novel architecture is based only on loops and directional couplers and does not include intersecting or branching sections, which notoriously increase the overall loss of an integrated optical circuit. Compared to other planar solutions reported in literature, such as those based on integrated beam splitter/combiners or gratings, the present architecture does not present counter directional recirculation, and does not suffer from in-plane scattering and diffraction limitation, being based on channel waveguides. Moreover, because of an extra zero-pole in its transfer function, it results more flexible in a design of appropriate spectral responses for filtering operations. As an application of the presented architecture a Toeplitz matrix-vector multiplier is presented.

In this paper we show how the Fresnel transform helps to detect time-delayed short pulse signals. A short pulse disperses spatially as it propagates through free space according to the Fresnel transform. A point photodetector produces a signal whose temporal frequency is directly proportional to the time delay between the pulses. When this signal is fed to an acousto-optic spectrum analyzer, the frequency domain represents a time delay domain.

Five channel direction of arrival processor is shown using TeO₂ shear wave, five channel acoustooptic Bragg cells, and operating in the 30 - 90 MHz frequency range. Description of the complete system is given including the method to eliminate the bidirectional characteristics of the linear antenna arrays in order to cover the full 0 - 360 degree(s) range.

Optical signal processing architectures have been designed which offer potential solutions to Air Defense Initiative (ADI) radar signal processing requirements for low radar cross-section target detection and noncooperative target recognition (NCTR). A rack-mounted engineering development model of a wide-bandwidth acousto-optic correlator has been developed to provide high range resolution for target feature extraction and discrimination. A wideband acousto-optic range-Doppler processor brassboard has also been designed to provide simultaneous high range resolution and Doppler filtering. Design and fabrication of a test radar subsystem is in progress, and will be utilized as an ADI clutter suppression test-bed for integration and testing of the high resolution acousto-optic correlator engineering development model and the acousto-optic range-Doppler processor brassboard. In this paper, the designs and status of the optical processors and test-bed are described.

Two acousto-optic devices are described which, in contrast to the more common Bragg Cell systems which process varying RF inputs by deflecting laser radiation, analyze broad optical spectral inputs using essentially constant RF inputs. The principles described in this paper have been applied to the design and manufacture of a Wide Angle Staring Spectrometer, which is also discussed.

There has been a number of significant advances in recent years in the conception and realization of new and novel guided-wave acoustooptic (AO) devices and related integration technologies in common waveguide substrates. In this paper, a selective review of the most recent advances in the realization of multi-channel integrated AO modules or circuits in LiNbO₃ and GaAs substrates with applications to communications, RF signal processing and computing is presented.

Some important results recently achieved with in-house grown high-quality, high-performance Hg₂Cl₂ and PbBr₂ are described. Using an optimized crystal growth technique we have grown Hg₂Cl₂ crystals that show a significantly reduced acoustic attenuation from 13.4 to 8 dB/microsecond(s) -GHz². These crystals allow the development of Hg₂Cl₂ Bragg cells with TBWP figures in the 5,100 - 6,900 range, frequency operation as high as TeO₂ and resolution about 123% higher than TeO₂ for practically similar crystal lengths. We have also grown for the first time long, high-quality PbBr₂ crystals which exhibit a large figure of merit (M₂ equals 550) with a modest attenuation figure of 12 dB/microsecond(s) -GHz². This material may be the choice for devices where large diffraction efficiencies are needed.

The phased array transducer and birefringent Bragg interaction provide a unique combination of features not available with the usual acoustical or optical rotation schemes in slow shear TeO₂. A general theory for phased array Bragg interaction in a birefringent medium is presented. A 50 MHz bandwidth slow shear TeO₂ Bragg cell was constructed using this theory. Excellent match between the designed and observed frequency band was obtained. As a result of the strong anisotropy of the acoustic velocity, we were able to reduce the intermodulation products by removing the negative first-order acoustic beam from the optical interaction region. The limitation in performance imposed by previous slow shear TeO₂ designs using either optical or acoustical rotation are eliminated.

The advanced development and characteristics of a new high resolution, high frame rate, reflected mode MOSLM is reported. This effort is aimed at the production of Miniature Ruggedized Optical Correlators. The device research and process development is being performed at Carnegie Mellon University NSF Data Storage System Center (formerly Magnetic Technology Center) under contract from Litton Data Systems. Pixel size is under one mil center to center, one third the dimension of present transmission mode evaluation devices. This development includes optimization of the optical and functional characteristics of the MOSLM for Mil Spec Systems.

We described here an integrated spatial light modulator (SLM) employing Quantex electron trapping (ET) materials. The light modulation is accomplished by emission of ET material, upon incident coherent infrared light, where a pattern is written to by previous visible light excitation. The ET based spatial light modulators offer unique advantages over other SLM devices, such as capability of converting incoherent input to coherent light output and of integrating the modulator, the photodetector, and the memory into a single, rugged unit.

It is well known that real-time pattern recognition and tracking can be accomplished with optical correlation. The performance of a particular correlator depends mainly on the capability and availability of real-time spatial light modulators. Recently, twisted nematic liquid crystal (TNLC) devices have been used as spatial light modulators. To date, most of the correlator systems use these liquid crystal devices in either a joint transform correlator or as the input spatial light modulator in a Vander Lugt correlator system. This paper discusses the design considerations of a gray-scale converging beam correlator system using liquid crystal devices at both the input and the filter planes. The optical properties of twisted nematic liquid crystals pertaining to their use as a SLM are discussed. To date, all of the pixelated liquid crystal devices are addressed by video signal. A gray scale video addressing scheme is presented. In addition, the design of a real time converging beam correlator utilizing liquid crystal devices from an Epson Video projector is presented along with results that characterize its performance.

We have constructed a correlator using a gray scale amplitude modulator for an input device and a gray scale phase modulator in the filter plane. The spatial light modulators used in the correlator are two of the three liquid crystal cells from a commercially available color projection television. The cells (twisted nematic liquid crystals) may be operated in an amplitude mostly or phase mostly mode by selecting the polarization of the light and the operating bias voltage. We show the Jones matrix analysis of the laboratory measurements and the corresponding operating curves. Signal to noise ratios on the order of 100:1 have been observed for some test objects using simple phase mostly filters.

An optical architecture based on optically addressed amorphous silicon ferroelectric liquid crystal spatial light modulators has been built and demonstrated for optical iterative processing of two dimensional images. An optical ring in which a two dimensional optical delay line has been introduced, allows control of the cycling rate of the optical iterative processor from a few Hertz to up to 15 kHz. Potential applications for real time optical iterative processing includes scaling and rotation for optical based ATRs, optical fractal analysis, optical wavelet transform, and optical artificial neural network implementations.

With the growing interest in using binary phase only filters (BPOF) in optical correlators that are implemented on magnetooptic spatial light modulators, an understanding of the effect of errors in system alignment and optical components is critical in obtaining optimal system performance. We present simulations of optical correlator performance degradation in the presence of eight errors. We break these eight errors into three groups: 1) alignment errors, 2) errors due to a combination of component imperfections and alignment errors, and 3) errors which result solely from non-ideal components. Under the first group, we simulate errors in the distance from the object to the first principle plane of the transform lens, the distance from the second principle plane of the transform lens to the filter plane, and rotational misalignment of the input mask with the filter mask. Next we consider errors which result from a combination of alignment and component imperfections. These include errors in the transform lens, the phase compensation lens, and the inverse Fourier transform lens. Lastly we have the component errors resulting from the choice of spatial light modulator. These include contrast error and phase errors caused by the non-uniform flatness of the masks. The effects of each individual error are discussed, and the result of combining all eight errors under assumptions of reasonable tolerances and system parameters is also presented. Conclusions are drawn as to which tolerances are most critical for optimal system performance.

In the image analysis via the general theory of moments, some shortcomings were discovered. The major disadvantages of the method of moments are that although the first few moments convey significant information for simple objects, they fail to do so for more complicated objects. The computed invariant moments are expected not to be strictly invariant in the presence of noise, and the computation of moment invariants is time-consuming. In this paper, an effective image analysis method based on the algebraic method is presented. First, a new coordinate transformation from an object image to an invariant matrix is proposed. The invariant matrix representation is independent of image translation, scaling, and rotation. On the basis of invariant matrix, a robust algebraic recognition method based on the projective image is developed. The projective coordinates of projective image on feature images are used as the feature vectors which represent the inherent attributes of objects. In order to test the efficiency of our method, it is used to solve the recognition problem of two-dimensional aircraft models. Theoretical and experimental results show that our method is very reliable for image analysis and the extracted algebraic features are invariant to image translation, scaling, and rotation.

We present a new architecture of an adaptive optical processor, constructed according to adaptive learning and adaptive resonance architecture. It has four main properties.

Centroid and geometric center methods are suggested for tracing the center location of an irregular interferogram in this paper. The methods remove limitations of general software which may only analyze straight or quasi-straight fringe patterns. The F-test method is used to control accuracy of fringe analysis. The speed and accuracy of the software depend upon the quality of the fringe pattern and a number of sample points.

The history of acousto-optic is discussed. Numerous architectures of acousto-optic systems for spectral and correlation processing of electric and optical signals are analyzed. The results of experimental studies of signal processors and the parameters of experimental prototypes of acousto-optic cell prototypes are presented.

A review of the papers in the field of acousto-optics carried out at the Tomsk institute of automatic control systems and radioelectronics from 1969 to the present is presented here. The results of fundamental investigations concerning ultrasound excitation in piezocrystals and the acousto-optic interaction in anisotropic media are given. The problems of wideband HF Bragg cells are also considered. The principles of construction, theoretical models, and main characteristics, constructions, and results of experimental and natural investigations of acousto-optic processors of radio signals (receivers-frequency meters, phase-meters-frequency meters) are described as well.

There are several research centers in Russia which deal with optical information processing (OIP). They are primarily located in Moscow: the P. N. Lebedev Physics Institute and the Institute of Information Transmission Problems, belonging to the Russian Academy of Sciences, the Institute of Optical and Physical Measurements of the Russian Committee of Standards, Moscow State University and Moscow Physical Engineering Institute. Some of the advances of these centers in OIP will be mentioned, as well as those of the Institute of Automation and Electrometry of Sibirean branch of the Russian Academy of Sciences (Novosibirsk-city).

Results of the multichannel electro-optic waveguide modulator investigation are presented. The possibility of modulator applications for recognition of analog and digital signals is shown. Influence of channel nonidentity on process of data input by the modulators is analyzed.

The paper considers the possibility of using acousto-optic devices for compensation of interstellar dispersion influence in observations of radio emission from pulsars. The principle of operation in an adaptive acousto-optical processor using a CCD-line photodetector in the shift-and-add mode is discussed. An adaption is fulfilled by control CCD scan rate. Results are reported for experimental investigations in the laboratory breadboard setup.

The results of investigations on acousto-optic interaction of quasimonochromatic fields in crystals are presented. Light diffraction in waveguide modulators have been considered, and light beam diffraction in thermo-disturbed crystals has been investigated.

An optical system was constructed and tested that reads data in word parallel form from a selected address in an optical random access memory (RAM), and passes it to selected addresses in another RAM. The advantage of using optical random access memory is that software developed for a sequential electronic machine is easily transferred to this optically enhanced machine because the same instruction set may be used. In early optical computers, this advantage may outweigh not using optics to its full parallelism potential. An bit-slice search technique is used to generate a beam whose level is dependent on an optical address selected. Bit-slice units are used to select the read and write addresses. The system moves the data from a selected address on one storage device to one or more selected addresses on another. An optical experiment, using optically addressable liquid crystal light valves, image intensifiers, and spatial light rebroadcasters (SLRs), shows how a row of data may be selected optically for reading and then fanned out to more than one location specified by separate optical addresses.

Si systems are rapidly evolving towards more efficient, compact parallel architectures characterized by a mix of both monolithic and hybrid technologies as silicon technology moves into the ultra large scale integration (ULSI) era. Previous distinctions between active chips and passive packaging will disappear within multi-chip modules (MCM) in which the Si interconnection substrate has evolved active devices supporting an interconnection network linking components within the module. As architectures emerge based upon multiple MCMs, the role optics plays as a performance enabling technology will be determined by how well it can provide a high performance interconnection network function which can efficiently bridge the physical substrate and package boundaries of the system. This paper motivates work experimentally exploring the compatibility of GaAs heteroepitaxy with submicron Si CMOS and the scaling of polymer high density optical waveguide arrays in order to achieve high density MCM-to-MCM optical interconnections compatible with emerging MCM environments.

Routing performance of optical interconnection networks is limited by the complexity of switches and the connectivity of the networks. One way to overcome these limitations is to allocate the network bandwidth in a time-division multiplexed (TDM) way. More specifically, an appropriate subset of input-to-output connections can be established during a time slot and all possible connections are established within several time slots. That is, the network is reconfigured with time division multiplexing at an appropriate degree to emulate a fully connected network. Message routing can be done by selecting an appropriate time slot in which the required connection is established. However, the connection latency, which is equal to the multiplexing degree, may be prohibitive in a large network. To reduce the latency, only a subset of all possible connections needs to be established in the network with time-division multiplexing as required by applications. Network reconfiguration with TDM may be done either statically or dynamically. Static reconfiguration can be based on compile time analysis of an application program, while dynamic reconfiguration is controlled at run time. With time- multiplexing, several virtual networks are created in the time domain and the control overhead can be amortized over the number of virtual networks. Simulation studies have been carried out and results show that dynamic reconfiguration with TDM can effectively ease the communication bottlenecks.

Convolution of long sequences of data is often needed for various sensing, signal processing, and pattern recognition applications. In this paper, massively parallel opto-electronic interconnect schemes are proposed to solve the problem of real-time convolution of long (10⁴ - 10⁶) data systems. Based on the Chinese remainder theorem, a 1D data sequence of length N (N equals r₁r₂, where r₁ and r₂ are mutually prime to each other) may be permuted into a 2D array of a size r₁ by r₂. A convolution of the two 1D data sequences each of N points is therefore converted into a convolution of the two corresponding 2D arrays each of a size r₁ X r₂. A standard 2D optical image convolver sandwiched between an input and an output opto-electronic data permutation devices (interconnect networks) can thus perform the required convolution. Two video rate opto- electronic data permutation schemes which are based on: (1) the use of a modified cathode ray tube (CRT) and (2) a combination of a standard CRT and an optical geometric transformer, respectively, are described. The permuted 2D data are subject to a standard free-space optical convolution before a 2D to 1D inverse permutation to generate the final 1D convolution result. Computer simulation for the entire three-stage algorithm and their results are discussed. Technical problems and fundamental limitations of the proposed schemes are also discussed.

Convolution of long sequences of data is often needed for various sensing, signal processing, and pattern recognition applications. In this paper, massively parallel opto-electronic interconnect schemes are proposed to solve the problem of real-time convolution of long (10⁴ - 10⁶) data systems. Based on the Chinese remainder theorem, a 1D data sequence of length N (N equals r₁r₂, where r₁ and r₂ are mutually prime to each other) may be permuted into a 2D array of a size r₁ by r₂. A convolution of the two 1D data sequences each of N points is therefore converted into a convolution of the two corresponding 2D arrays each of a size r₁ X r₂. A standard 2D optical image convolver sandwiched between an input and an output opto-electronic data permutation devices (interconnect networks) can thus perform the required convolution. Two video rate opto- electronic data permutation schemes which are based on: (1) the use of a modified cathode ray tube (CRT) and (2) a combination of a standard CRT and an optical geometric transformer, respectively, are described. The permuted 2D data are subject to a standard free-space optical convolution before a 2D to 1D inverse permutation to generate the final 1D convolution result. Computer simulation for the entire three-stage algorithm and their results are discussed. Technical problems and fundamental limitations of the proposed schemes are also discussed.

In this paper, we study the constructions and mathematical expressions of butterfly interconnections and networks. Because 2-D and 3-D networks have different constructions, we discuss the transform from 3-D forms to 2-D ones. We start with optical Fourier transforms to derive the discrete transforms of 2-D and 3-D butterfly interconnect networks for implementing 1-D and 2-D fast Fourier transforms, respectively, which are the basic operations of intelligent optical computings.

Low-threshold, high-performance dry-etched ridge waveguide lasers with dry etched facets are of interest for monolithic two-dimensional coherent applications such as optical interconnects and optoelectronic integrated circuits. We report on low threshold current and wavelength emission < 8000 angstroms laser diodes with short cavity and dry etched facets. The facets are fabricated by reactive ion etching, which provides nearly vertical walls. For the first time, coherent GaAlAs/GaAlAs laser diodes (emission wavelength 7940 angstroms at room temperature) with cw threshold currents as low as 4 mA (room temperature) and 0.8 mA (at 77 degree(s)K) were achieved on a 4-micrometers -wide, 100-micrometers -long device.

The gain characteristics of strained layer InGaAs quantum well laser amplifiers grown on GaAs substrates have been characterized using a Ti-sapphire laser in both CW and self-mode- locked configuration. All-optical bistability was achieved in such a device through the nonlinear refractive index that arises as a result of the gain saturation of the diode amplifier at high optical intensities. The Ti-sapphire laser was modified to produce wavelength tunable pulses of the order of 150 fs with a 15 nm spectral bandwidth in order to measure the dynamics of the gain using a time resolved optical pump-probe technique. The gain saturation occurs on the time scale of the pulse width of the laser and the initial recovery is extremely fast. This initial speedy recovery on a time scale of 100 fs is followed by a slower recovery with a time constant of 3 ps followed by a much slower recovery of 500 ps.

Analog testing and primitive evaluation has been performed on OptiComp's second generation digital optical computer (DOC II) platform. Preliminary results indicate that analog addition can be achieved at up to 6 bits of accuracy. The DOC II components which have been integrated for this study include: the laser diode array, the laser collimator and 1 X 1 imaging barrel, the point source modulator GaP Bragg cell, and a single channel APD/transimpedance amplifier. Two analog pulse trains are used to drive two independent sets of Bragg cell channels. The first order deflected light is collected and summed on an APD photodetector, amplified with a high gain, wide bandwidth, transimpedance amplifier and digitized using an Hewlett Packard digitizing scope. Testing was performed using a GPIB instrumentation interface. Ultimately, this analog computing architecture could produce a 256 vector multiply to 256 X 256 matrix which would yield a product with 8 bit resolution.

A fully parallel optical matrix-matrix multiplier is presented in this paper. Two N X N matrices are encoded on two tightly cascaded spatial light modulators (SLMs). One of the SLMs can be replaced by a light source array and a group of cylindrical lenses. Optical summation of all the outer product terms is performed by N lenses, which are arranged into a two-dimensional array to accommodate to the configuration of the SLMs used. Submatrices displayed on the SLMs are imaged by the corresponding lens in the lens array onto the output plane. Images of all the submatrices are superimposed and their intensities are added at the output plane to produce the multiplication result.

Problem preprocessing before insertion into a Bimodal Optical Computer (BOC) for solving Ax X b is shown to be required in some cases and desirable in all cases. With enough preprocessing we can achieve wonderful results (equivalent to inverting A). Unfortunately, this defeats the BOC purpose of fast solution. WE show that there are very simple, computationally-inexpensive preprocessing steps which solve some of the representation and conditioning problems, although no 'optimum' procedure is known.

A novel parallel model of natural language (NL) understanding is presented which can realize high levels of semantic abstraction, and is designed for implementation on synchronous SIMD architectures and optical processors. Theory is expressed in terms of the Image Algebra (IA), a rigorous, concise, inherently parallel notation which unifies the design, analysis, and implementation of image processing algorithms. The IA has been implemented on numerous parallel architectures, and IA preprocessors and interpreters are available for the FORTRAN and Ada languages. In a previous study, we demonstrated the utility of IA for mapping MEA- conformable (Multiple Execution Array) algorithms to optical architectures. In this study, we extend our previous theory to map serial parsing algorithms to the synchronous SIMD paradigm. We initially derive a two-dimensional image that is based upon the adjacency matrix of a semantic graph. Via IA template mappings, the operations of bottom-up parsing, semantic disambiguation, and referential resolution are implemented as image-processing operations upon the adjacency matrix. Pixel-level operations are constrained to Hadamard addition and multiplication, thresholding, and row/column summation, which are available in magnitude-only optics. Assuming high parallelism in the parse rule base, the parsing of n input symbols with a grammar consisting of M rules of arity H, on an N-processor architecture, could exhibit time complexity of T(n) <EQ O(MHn/N). When N equals O (Mn), which is feasible with massive parallelism, the computational cost is constant and of order H. Since H < < n is typical, we claim a fundamental complexity advantage over the current O(n) theoretical time limit of MIMD parsing architectures. Additionally, we show that inference over a semantic net is achievable is parallel in O(m) time, where m corresponds to the depth of the search tree. Results are evaluated in terms of computational cost on SISD and SIMD processors, with discussion of implementation on electro-optic architectures.

This paper will review the results of an 18 month program which developed and demonstrated a compact size general purpose Holographic-Based Digital Optical Processor, HBDOP. The HBDOP was developed to accommodate a number of different holograms. Three Holograms were developed; one performs parallel symbolic substitution for hetero-associative memory recall, a second performs 64 node hypercube interconnection network operation, and the third performs 64 node mesh-interconnected multiprocessor operation. The 64 node hypercube interconnect operation will not be discussed in this paper. The demonstrated systems utilize N^**4 holographic array recording to achieve massive interconnection parallelism not achievable with VLSI technology. Significant results of the program included the demonstration of 64 digital optical processors operating in parallel to perform digital operations such as morphological functions, simple target tracking, and noise removal operations using simple logic functions while simultaneously communicating through a mesh interconnection network. The demonstration of a 32 X 32 hologram array containing inter-pattern associations to act as an interconnection weight matrix for symbolic substitution and associative memory recall operations. The demonstration of a 64-node hypercube interconnection network by using the N^**4 hologram array to achieve 3-D parallel interconnection. Finally, the development of an automatic system for N^**4 holographic array recording. The automatic recording system is capable of recording one hologram element in 10 seconds.

A hybrid electro-optical computer has been constructed. The computer consists of an electronic host computer and a digital optical coprocessor. The optical coprocessor is a free space implementation of an optical programmable logic array. System architecture and fabrication are discussed.

Optical processors are attractive because of their ability to perform massively parallel operations such as matrix vector products. The inherently analog nature of optical calculations requires that optical processors be based on analog computations. While the speed at which such analog operations can be performed as well as the natural parallelism of optical systems are great advantages of optical processors, the analog representation of values severely limits the achievable accuracy. Furthermore, optical processors are limited by the need to convert information to and from the intensity of light. Digitization can be used to increase the accuracy of optical matrix-vector processors, but causes a severe reduction in speed. This paper compares the throughput and power requirements of optical and electronic processors, showing that optical matrix-vector processors can provide a greater number of operations/Watt than conventional electronics.

Image segmentation by texture is achieved by using Gaussian smoothed fan (GSF) filters in the Fourier domain. These filters have intrinsically more defined bandwidths than Gabor filters and were developed earlier by the authors for multichannel filtering on digitized images. This paper reports a Fourier optical implementation of GSF filters on images constructed from Brodatz album textures. The back-transformed images are compared with corresponding output images obtained from digital processing and demonstrate that the approach is well suited to real-time fast segmentation of images containing texture fields.

An acousto-optic processor capable of analyzing signals consisting of a high frequency carrier modulated by an envelope signal is described. Space-integrating spectral analysis is used to channelize signals by carrier frequency. Time-integrating spectral analysis is used to characterize the envelope signal that modulates each carrier frequency. The output is a two- dimensional display with carrier frequency along one axis and envelope frequency along the orthogonal axis. Several advantages of the processor are explained and proof-of-concept experimental results are presented. One possible application of the processor is to the automatic separation and determination of the carrier frequencies and pulse repetition frequencies of multiple received radar signals.

In this study, the algorithm is described for investigation of spatial speckle-noise characteristics such as a common area occupied by speckles, a distribution probability function of spot areas, linear sizes, etc. The main features of the processing algorithm is to pick out a large number of separate speckles from a whole image.