High data rate optical computing is a major emerging technology, driven by an ever-increasing need for the processing of information. The current volume of technical literature on this subject attests to its new prominence as a computing technology complementary to digital electronic computers. The papers in this special issue of Optical Engineering, together with those published in a recent IEEE Proceedings (July 1984), represent a comprehensive cross section of current research interests in optical computing.

Devices for optical processing and computing systems are discussed, with emphasis on the materials requirements imposed by functional constraints. Generalized optical processing and computing systems are described in order to identify principal categories of requisite components for complete system implementation. Three principal device categories are selected for analysis in some detail: spatial light modulators, volume holographic optical elements, and bistable optical devices. The implications for optical processing and computing systems of the materials requirements identified for these device categories are described, and directions for future research are proposed.

Optical systems are particularly adept at performing high speed multiplication and division, which are the operations required for matrix multiplication. Implementing these operations in an integrated optical circuit (IOC) has the additional advantage of compactness and manufacturability. In this paper it is shown how analog integrated optical components can be assembled to design compact 10Cs for implementing both analog and digital algorithms for carrying out matrix multiplication.

Recent progress in coherent optical pattern recognition is reviewed. Emphasis is given to techniques for multiclass distortion-invariant pattern recognition. Real-time and practical architectures and algorithms are described, and recent results on extensive data bases are noted. Feature-extraction and correlation architectures are chosen as the two major approaches. The feature-extraction techniques considered include Fourier coefficients, chord distributions, and moments. The correlator approaches emphasize the use of synthetic discriminant functions to achieve multiclass distortion-invariant pattern recognition.

A review of optical range sensing techniques for machine vision is given. Four basic categories of range sensing techniques are discussed: geometric techniques, time-of-flight techniques, interferometric techniques, and diffraction techniques. The basic principles are elucidated, and general comparisons are made between the groups. Representative examples are given of many different approaches. The challenge for optics and optical computing is to develop new range sensors that are fast and accurate and require little or no post-detection processing.

Confocal feedback systems exist in a variety of forms and can solve a wide range of partial differential equations (PDEs) and integral equations (lEs). In this paper we describe several of these feedback systems and how they can be applied to provide optical analog solutions to PDEs of constant coefficients (e.g., diffusion, Poisson's, and wave equations), PDEs of variable coefficients (e.g., modified Helmholtz equations), three-dimensional PDEs, four-dimen-sional PDEs, and IEs (e.g., Fredholm and Volterra equations). The important advantage of obtaining the solutions by optical analog methods rather than digital methods is speed. The disadvantage is solution accuracy, although the accuracy obtainable with optical feedback is better than without feedback. To further improve solution accuracy, we suggest the replacement of simple spherical mirrors by Mangin mirrors and the incorporation of coherent image amplification by photorefractive crystals (e.g., BaTiO3 or BSO) in the confocal systems.

The increasing need for large-scale computations has led to new interest in optical parallel computing. Digital parallel processing can be implemented using optical truth table look-up techniques. With optical-logic-based pattern recognition, a content-addressable memory can be constructed. The use of the EXCLUSIVE OR and NAND logic operations to achieve content addressability is discussed. This memory system can be used to perform digital truth table look-up processing. Operations such as addition and multiplication of 4-, 8-, 12-, and 16-bit words in parallel arrays are then directly possible. The number of reference patterns that must be stored is dramatically reduced by the use of the (binary-coded) residue number system and logical reduction tech-niques. An example 16-bit multiple-word-parallel (of order 1000) fixed-point multiplier is discussed.

Optical systolic processors perform rapid parallel multiplications of vector and matrix elements. These multiplications may be performed by convolving the digits of individual numbers. Current transducers (particularly acousto-optic cells) make time-domain convolution the method of choice. A time-domain convolver can operate in time-integrating or space-integrating modes. These two modes can be considered as variations of a basic multiplier unit. Starting with the basic multiplier, four matrix-vector and two matrix-matrix multipliers can be derived. Some of these processors are new, and some have been previously identified. Within this unifying framework the perfor-mance of all these processors may be analyzed, and design trade-offs may be identified.

Defining a channel failure in an optical systolic array processor leads to a simple test for failure and an adjustment procedure that will correct minor failures. Noncorrectable, or major, failures must be bypassed when possible or limited in effect when no further bypassing is possible. This analysis leads to the concept of hierarchies of modules in which failures can be detected and either healed or bypassed. Software to detect instantaneous failures is described. Such a detection triggers additional correction by bypassing.

In this paper we present the basic principles of optical bistability and summarize the current advances in semiconductor optical switching, with emphasis on recent results in GaAs, CuCI, InAs, InSb, CdS, ZnS, and ZnSe etalons. These devices have great potential for applications involving optical signal processing and computing. As an example, we discuss the use of arrays of bistable devices for parallel optical processing and for addressable spatial light modulators. The use of nonlinear etalons as optical gates is also illustrated. To date, GaAs devices have shown the most favorable characteristics for practical applications. They operate at room temperature with a few milliwatts of power using a laser diode as the only light source. Quasi-cw operation and optical fiber signal regeneration have also been demonstrated. A GaAs NOR gate operates in 1 ps with <3 pJ incident energy; this, of course, implies a 1 ps switch-on time for a bistable etalon.

In this paper we present the design for an optical switching system for minicomputers that uses an optical spatial light modulator such as a Hughes liquid crystal light valve. The switching system is designed to connect 80 minicomputers coupled to the switching system by optical fibers. The system has two major parts: the connection system that connects the data lines by which the computers communicate via a two-dimensional optical matrix array and the control system that controls which computers are connected. We first present the basic system, then describe the matrix-based connecting system and review some of the optical components to be used. Finally, the details of the control system are given and illustrated with a discussion of timing.

The well-known central-slice, or projection-slice, theorem states that the Radon transform can be used to reduce a two-dimensional Fourier transform to a series of one-dimensional Fourier transforms. In this paper we describe a practical system for implementing this theorem. The Radon transform is carried out with a rotating prism and a flying-line scanner, while the one-dimensional Fourier transforms are performed with surface acoustic wave filters. Both real and imaginary parts of the complex Fourier transform can be obtained. A method of displaying the two-dimensional Fourier transforms is described, and representative transforms are shown. Application of this approach to Labeyrie speckle interferometry is demonstrated.

At low light levels, detected photoevents can provide a physical source of random numbers. In turn, these optically generated random numbers can be used to construct Markov chains, which form the mathematical frame-work for many of the Monte Carlo procedures. In this paper a method for optically generating bivariate random deviates with a specified probability den-sity function and correlation coefficient is given. The optical implementations of two Monte Carlo procedures also are discussed. First, a method to calculate two-dimensional definite integrals is considered. Next, an optical method for solving systems of linear algebraic equations by the Von Neumann-U lam Monte Carlo procedure is given.

This special issue of Optical Engineering addresses a number of critical issues in the continuing invention, development, and characterization of components for optical information processing and computing applications. This is the second in an annual series of related special issues, initiated with the publication of "Spatial Light Modulators: Critical Issues" (Optical Engineering, Vol. 22, No. 6, November/ December 1983). Although the last two decades have witnessed tremendous progress in the invention and development of optical processing techniques, architectures, and algorithms, only limited optical hardware is currently available for the implementation of such concepts. As a result, extensive software and numerous system schematics exist for a multiplicity of processor configurations that in almost all cases have not proven implementable in compact systems capable of high frame rate operation. A notable exception is the excellent performance record of optical processors based on one-dimensional acoustooptic transducers. Such acoustooptic devices have gained wide use and acceptance in military and commercial sectors for applications such as spectrum analysis, signal correlation, light modulation, and laser beam scanning and steering.

The third in the series of special issues of Optical Engineering addressing significant advances in materials and devices for optical information processing has been scheduled for publication in January/February 1986.

An alternative to the conventional flying-spot scanner architecture is the Scophony scanner. The Scophony scanner uses the same optical elements as the more familiar flying-spot scanner: a rotating polygon mirror, an acousto-optic (A/O) modulator, and a laser light source. The flying-spot scanner is designed to construct its image a pixel at a time; no more than one pixel is illuminated at any given instant. The Scophony scanner is designed to image a broad swath of the A/0 modulator's acoustic pulses onto the photoreceptor. Many pixels are illuminated at any given instant in the Scophony scanner. The result is a scanner with a coherent imaging response. This coherent response implies that the phase of the modulator's electronic drive signal for a given pixel profoundly influences the formation of the neighboring pixels at the scanner image plane. This coherent response enables electronic manipulation of the video drive signal to have significant impact on the optical imaging performance of the scanner. In this paper, two electronic manipulation schemes are pro-posed for doubling the resolution of the Scophony scanner, one scheme for analog video signals and one scheme for binary digital video signals. Each scheme gives superior contrast ratio performance when compared with the flying-spot scanner.

Electromechanical devices can be used to rapidly modulate light if their mechanical inertias can be made small enough. The usefulness of these devices depends upon the ability to produce them in large arrays with high yield and at low cost, to integrate them with their drive circuitry, and to achieve a high level of reliability and reproducibility. In recent years micromechanical fabrication techniques have been used to produce arrays of miniature optical modula-tors on silicon that can operate at frequencies to 1 MHz with negligible drive power. They are completely compatible with integrated circuit technology and can be fabricated on a silicon wafer along with electronic circuits. In large arrays they can transfer data at a high rate for parallel optical readout of microelectronic devices and sensor arrays. Once in optical form, the data are amenable to optical processing. The development of the micromechanical modulator, its features, performance characteristics and limitations, and its potential applications to data transfer and processing are discussed.

The architecture, operation, and performance of a new two-dimensional, optically addressed, membrane spatial light modulator are described. The modulator, the optical-to-optical deformable mirror device, consists of an array of 128 X128 deformable mirror elements on 25 .im centers integrated on silicon. The device exhibits a sensitivity of 1.7 jJ/cm2 at 525 nm, write times of 25 i. is, and erase times of 40 I.Ls. Images are read out with near-IR illumination. The device can be used to convert incoherent images to coherent images for optical information processing applications.

The development of a liquid-crystal-based visible-to-IR dynamic image converter (VIDIC) is described. Liquid crystal studies in the IR as well as the structure, operation, and preliminary performance results of a 10.6 um VIDIC are reported.

A vacuum-demountable prototype electron-beam-addressed spatial light modulator that consists essentially of an electron gun, a microchannel plate, and an electro-optic crystal is being investigated. This paper discusses the underlying principles of operation of the device, which include the interaction of an electron beam with a microchannel plate and the addressing of an electro-optic crystal with a flux of nonmonoenergetic electrons.

Volume holographic storage in photorefractive Bit 2S i020 (BSO) crystals is utilized to perform dynamic incoherent-to-coherent image conversion by means of selective spatial erasure of a uniform grating with white (or quasi-monochromatic) light. Physical characterization of device performance is de-scribed, with emphasis on resolution, sensitivity, and temporal response. The photorefractive incoherent-to-coherent optical conversion process is analyzed in terms of a simple model that relates the diffracted intensity to the space-variant effective modulation ratio.

This paper discusses the materials issues of acousto-optic (AO) devices. Criteria for the selection of materials for the three basic types of AO devices, deflectors (Bragg cells), modulators, and tunable filters, are presented. Comparison of physical properties of the various AO materials shows that trade-offs are required to suit specific applications. As an example of how to search for new materials, the potential of a chalcopyrite compound (ZnGeP2) for AO device applications is discussed. The paper includes tables that list the properties of selected AO materials.

Compositions of lithium niobate containing 4.5 at.% or more MgO have the ability to transmit, without distortion, light 100 times as intense as undopecl compositions. Holographic diffraction measurements of photorefraction have demonstrated that the improved performance is due to a hundredfold increase in the photoconductivity, rather than to a decrease in the Glass current. The damage-resistant compositions are also distinguished by a thermal activation energy of 0.1 eV for the diffraction efficiency, an OH-stretch vibration at 2.83 Am, a lattice phonon absorption at 21.2 Am, an electron spin resonance signal for Fe impurities at 1500 G, and, after reduction by heating in a vacuum, an optical absorption band at 1.2 um. (The corresponding values for undopedl LiNbO3 are 0.5 eV, 2.87 um, 21.8 um, 790 G, and 0.5 um, respectively.) The high photoconductivity is thus related to a distinctive electronic environment for impurities in the damage-resistant crystals. The photoconductivity strongly affects the impedance and time constants of signal processing devices made of LiNbO3.

A complete real-time cross-correlator using both a photoconductive bismuth silicon oxide [Bi12Si020 (BSO)] liquid crystal light valve for data input and a photorefractive BSO crystal as a dynamic holographic filter is presented. Some specific properties of this dynamic optical processor are reported. The following optimum operating conditions of the light valve were used in the experiment: driving voltages, 30 Vrms; frequency, 300 Hz; and writing energy of the light valve in the blue-green spectral range, 150 /uJ cm -2. For a 15µm thick pentylcyanobiphenyl (PCB) liquid crystal layer, the measured spatial resolution at 50% MTF is 12 1p mm-1, thus ensuring a potential processing capability of 300X:300 pixels. The optimized time constant of the operating optical processor, including both the light valve and the photorefractive BSO crystal, is around 50 ms for 0.2 mW cm-2 incident intensity on the light valve at the green line of an argon laser. Experimental cross-correlation results for various types of digital or analog data written on the light valve are shown in this paper. Compensation of the residual distortions of the incident signal beam emerging from the light valve has been achieved by generation of a phase conjugate wavefront diffracted by a high efficiency holographic lens recorded on dichrornated gelatin.

We suggest and demonstrate a simple grating projection technique for slope and deformation mapping of large diffusive objects. The method is an extension of the multislit lichtschnittverfahren technique.

Starting with the spinor representation of the electromagnetic field, we consider the propagation of a light beam along the Oz axis of a nonlinear medium. We first prove that for nonstationary processes in a one-dimensional space, the energy conservation takes the form of a conservation equation of kinematic waves. Then, using the method of characteristics, we discuss the self-steepening problem for initial value and boundary condition problems. We give the shock conditions and the shock velocity. The paper ends with a discus-sion of self-modulation.

All imager arrangements degrade gray value pictures by their photo-response nonuniformity. This paper describes an adequate correction procedure. It is applicable in the case of rapidly changing image sequences containing a homogeneous background. As such systems are of interest in real-time applications, a detailed hardware design, including an error analysis, is given. Finally, the sources of nonuniformity are localized and illustrated by numerical examples.

Afocal cylindrical lens systems and their prismatic equivalents are used in the formation of multiplex holograms. They introduce the desired, controlled astigmatism, and they assist the low f number final cylindrical lens in the formation of a strip hologram. In addition, a line source and a broad source fringe system applied to the hologram-making process can result in improved signal-to-noise ratios.

Several configurations of charge-coupled device (CCD) imagers formed on gallium arsenide are considered. A process was investigated for the fabrication of overlapping charge transfer electrode structures in CCDs on GaAs with ion-implanted active channels. The electrode metal was aluminum and the interelectrode isolation medium was anodically formed aluminum oxide. The use of aluminum electrodes and ion-implanted active channels makes this fabrication process compatible with that used for fabricating high speed integrated circuits on GaAs. Support circuitry can thus readily be integrated with the CCD imager. A computer model that simulates charge transfer through an implanted channel beneath an array of electrodes with a finite gap between each pair of electrodes is presented. The use of this model in designing the overlapping electrode structure and channel profile is demonstrated. The impetus of the device design is to maximize charge transfer efficiency and minimize channel noise.

For the special case of monolithic ranging telescopes, a family of characteristic design curves for various system performance requirements is discussed. The required secondary mirror displacement and the tolerance on this position, as well as the corresponding depth of field for any desired range, can be readily obtained from this normalized family of curves. The hyperfocal distance (closest range within depth of field when infinite range is also in focus) is also displayed on this set of curves and can be used to determine when it is necessary to activate ranging. The degradation in system performance is then plotted versus range, and the closest effective range is determined. This range is a strong function of telescope diameter and is crucial to certain ranging telescope applications. This degradation in system performance is then deter-mined for various properly phased synthetic-aperture systems (multiple-mirror telescopes) and compared to the monolithic telescope of equivalent aperture. For certain applications these results provide a strong motivation for going to synthetic-aperture telescopes based upon optical performance alone.

An experimental data processing system has been developed to demonstrate the feasibility of processing high speed, multispectral image data on board the spacecraft. The design incorporates real-time processing with adaptable operation in an expandable architecture. The experimental hardware is coupled to test support and computing equipment to provide a laboratory tool for evaluating and demonstrating each of the processing functions as well as the overall system operation. An evaluation of the high speed processor was conducted to ensure that the desired system throughput was achieved without sacrificing processing accuracy. A description of the system is presented along with the results of the test and evaluation activity.

Recent advances in thin film fabrication techniques have made multilayer mirrors an important new addition in the field of soft x-ray diffraction optics. A multilayer is a one-dimensional periodic structure that consists of an alternating sequence of high and low atomic number elements evaporated or sputtered onto a substrate. Characterization of these devices has been carried out on a variety of x-ray sources, including synchrotron radiation, over a large range of parameters. Extensive modeling of the diffraction characteristics of multilayers has shown that the theory of optics of thin films gives good agreement between measured and calculated diffraction profiles. Application of multilayer mirrors as normal incidence x-ray collectors, monochromators, and beam splitters is being explored.

Multiple-reflection polarizers are designed using a single-reflection p- or s-suppressing polarizing mirror as the basic component, with the constraint that all mirrors be made of the same metallic substrate coated by the same dielectric film. Results are presented for polarizer designs using dielec-tric-coated titanium mirrors at the HF laser wavelength A = 2.8 um. The best designs are those that suppress the p polarization with all mirrors being coated by equal thicknesses of the same polarizing film. Excellent performance over a small operating wavelength range and in the presence of limited film-thickness and angle-of-incidence errors is demonstrated.

Oxide coatings for solar cells deposited from organic solutions were studied. Among them, an Al203 layer was found to decrease the surface recombination velocity of a silicon wafer to about 1 /10 of its value with no coating. The new constitutions were designed by inserting a thin Al203 layer between the silicon substrate and the antireflective layers. IR absorptions and the capacitance-voltage characteristics indicated that the reduced surface recombination velocity corresponded to the decrease of the surface state den-sity by the terminating action of H or OH. Instead of TiO2, a Ta205 layer was introduced to improve the absorption of the short wavelength radiation. By inserting the thin Al203 layer, the efficiency of the solar cell increased by about 5%.

This editorial is being written the day after the U.S. national elections. It is a time of change and yet a time of continuity. Likewise, in this journal there are both changes and assurances of continuity.