Optical systems for performing holography and optical processing are described. These systems are coherent, but operate with temporally incoherent light. The noise suppression mechanism of the system is analyzed. In particular, a method for deblurring an image is given.

We will describe a technique of optical information processing by a spectrally broad-band light source (e.g., white light). This technique enables the signal processing to be carried out in complex amplitude rather than in intensity. The basic advantage of the white light processing technique is the elimination of the coherent artifact noise that frequently plagues the coherent optical processing system. We will show that this white light processor is capable of performing complex signal filterings that a coherent optical processor can offer. We will demonstrate that signal detection by complex spatial filtering and image addition and subtraction can be carried out by this white light processing technique. Although this processing technique is effective in one dimension, however, it can be applied in 2-D problems in optical information processing.

The possibilities of wavelength multiplexing for integral operations are discussed. With different integration kernels we obtain encoding of position by wavelength, a correlation or a more general transformation. Most operations are performed only in one dimension. But two dimensional input functions (images) can be processed in parallel.

The application of polychromatic illumination and color-coded pupil apertures in optical transfer function (OTF) synthesis is discussed. Specific attention is focused on the use of photographic film in incoherent processors employing color-coding for OTF synthesis. Film is used primarily to provide a permanent record of an optical system output distribution but, under certain conditions, can also be used to perform image subtraction required in some synthesis operations.

The possibility of encoding information on an optical wave by modulating its temporal frequency (wave-length) and optically processing such data is investigated. A previously developed general formulation of optical processing systems is expanded to include a systematic exploration of possible wavelength diversity systems.

The use of white light in association with dispersive devices (prisms or gratings) leads to optical processing systems, specially correlators, that does not require hologram filters. Yet, the luminosity and the S/N ratio in the output plane are very high. We achieved a correlator from two spectroscopes in cascade. The first of them is illuminated by a spatially and temporally (white) incoherent object distribution, F(x,y), the different monochromatic images of which are shifted with respect each other. These shifted images scans the input transparency G(x,y) of the 2nd spectroscope. The convolution or correlation product ; F(x,y) * G(x/a,y) appears as an image at the output plane of the set-up (a denotes an arbitrary constant, depending on several geometrical parameters of the set-up). F(x,y) can be a real, self luminous, moving object and also G(x,y) can be a moving transparency (e.g cine film). One can use active components in the correlator (feedback). Examples are shown illustrating the ability of performing multichannel and space-variant processes.

The tunable acousto-optic filter has been described principally in terms of its capabilities as a rapid access spectrometer, and work has been reported on its multispectral imaging properties as well. Such filters are well adapted to microcomputer interfacing for control functions and hybrid optical-electronic operation. One of the most significant features of the device is the ability to rapidly scan a large spectral range, for example, the entire visible in less than 100 TJseconds, or to randomly access any of a number of selected wavelengths either singly or simultaneously. Applications have been reported in such fields as astronomical imaging and fiber optic systems, as well as the more conventional spectroscopic applications.

A single hollow cathode laser which emits twelve wavelengths at 3250, 4416, 5337, 5378, 6355, 6360, 7237, 7284, 8067, 8531, 8652, and 8878A from a mixture of He and Cd is described. Both single wavelength and simultaneous multi-wavelength cw laser action were obtained. The basic operating principle of this laser is described with particular emphasis on the various excitation mechanisms which lead to the realization of multi-spectral laser action. The laser tube employed has a modular design configuration with all metal and ceramic structure to ensure durable and reproducible operation. The typical laser performance under various He pressure, Cd density and excitation conditions is exemplified using the experimental results obtained from three-prime-color laser operation. From a 10 cm active length tube, the laser emits at least 10 mW in blue, 3.5 mW in green and 1.5 mW in red transitions individually. Simultaneous cw oscillation of these colors at total output power of about 30 mW is also demonstrated using a 25 cm active length tube. The laser power fluctuation is about 2 to 3 percent peak-to-peak with absence of discharge striation noise. Potential applications of this laser are briefly mentioned.

Two partially coherent and perpendicularly polarized vibrations Ex and Ey induce , in a polarization-sensitive photographic emulsion H , an anisotropy which is continuously varying across the plate . H , after exposure , is observed between crossed polarizers and reconstructs an interferogram of the vibrations whose contrast is maximum and whose black fringes localisation depends on the angle between the vibration incident on H and Ex .

An automated analytical electrophoresis microscope system was developed to provide instrumentation for rapid analysis of normal and altered cells in biologic and clinical studies. The system is controlled by a computer that (i) provides for the tracking of migrating cells, whose images are projected under phase contrast onto a vidicon, (ii) permits the rapid taking of multiple measurements per cell, and (iii) calculates the electrophoretic mobility. The system contains algorithms and software programs necessary to create a large data base of cell mobilities and to determine statistically valid distributions of cell mobilities.

Optical diffraction and spatial filtering methods have been used to determine the characteristics of periodic structures in many biological materials. The head shell of bacteriophage T4 has been chosen for this study, since aberrations in the assembly of the shell due to mutation or changes in growth conditions lead to the formation of a variety of elongated tubular head forms. The lattice parameters of structures assembled at elevated growth temperatures by normal, wild-type T4 and by a mutant (regA) have been analyzed using optical diffraction patterns obtained from electron micrographs. Spatial filtering procedures have been used for the reconstruction of one-sided images to determinE the characteristics of the head structures assembled under different growth conditions.

A gold-toning neutron-activation process has been developed at NRL for the purpose of intensifying or amplifying the contrast in extremely weak images in photographic negatives where the exposure level may be as low as 1.5% of optimum. Such contrast enhancement of photoreconnaissance and intelligence films allows the recovery of image information which may be taken under adverse lighting or obscured in the shadows and which might otherwise be lost by conventional methods using chemical/photographic or densitometer/computer techniques. The gold-toning neutron-activation process differs significantly from the usual nuclear methods of photographic image enhancement in two important ways: (a) A nonradiochemical toning solution is used to gold plate the image silver grains before activation. This procedure eliminates all hazards of handling radioactive solutions. (b) The neutron activation of the gold image is superior to either direct neutron activation of the silver image or radiochemical toning of the silver image both with regard to the efficiency of image activation and the speed and ease of producing autoradiographs. Recent results of the optimized gold-toning neutron-irradiation technique clearly demonstrated the recovery of image information which was not detectable by standard densitometer computer scans of the original film at the 1.5% exposure level. Such analog nuclear techniques can complement the power and flexibility of computer enhancement by significantly extending the lower limit of optical detection.

Utilization of both space and time integration in acousto optic signal processing over-comes some of the limitations of purely space integrating and purely time integrating techniques. A one-dimensional configuration for coherent hybrid spectrum analysis is used to show that this approach exhibits superior dynamic range characteristics, including a unique sidelobe suppression property. A coherent hybrid ambiguity function processor is discussed to illustrate the important advantages that may be exploited by using both space and time integration in delay-Doppler plane processing.

The paper discusses two novel applications of acousto-optics. In the first one, acousto-optics is used to read out the mode content of an acoustic cavity resonator, resulting in a two-dimensional spectrum analyzer. The second application deals with an electronically adressable optical filter that relies on Bragg diffraction imaging for its operation.

The problem of bias buildup in time-integration image processing is assessed. Numerical studies are presented that compare the results of the usual approach to interferometric time-integration processing with a proposed minimum-bias approach. It is concluded that the latter approach yields outputs of dynamic range adequate for many practical image processing situations.

A new architecture for performing time-integration correlation is described. The correlator uses a surface acoustic wave (SAW) delay line, and features the optical interference of two coherent light beams which have been Bragg-diffracted by SAW's propagating in the line. The time integration is performed by a photodiode array which detects the diffracted light. Time-bandwidth products exceeding 106 (50 MHz times 30 ms) have been achieved. This two-beam SAW acousto-optic time-integrating correlator has been used to detect a number of wideband spread-spectrum signals. It has several attributes which make it particularly well suited for use as a spread-spectrum signal processor. These include linearity of operation, large time aperture over which the correlation can be observed, and the ability to determine the center frequency and bandwidth of the signals. The suitability of this correlator for use as a signal processor in spread-spectrum systems is considered. In addition, a two-dimensional realization of this correlator is proposed for frequency scanning correlation. The use of this frequency scanning correlator as an LPI radar signal processor is discussed.

Optical systems capable of performing arithmetic operations in parallel on arrays of binary input data are discussed. Data obtained from the truth-tables for the implemented operations are stored as holograms used in the systems. These data (or their reduced form) correspond to either the sum of products or the product of sums logical expressions relating each of the bits of the result to the input bits. Using the operations of addition and multiplication as examples it is shown that the number of holograms that must be recorded is considerably reduced when the input and output data are in binary-coded residue representation rather than in direct binary representation.

A design for an optical general purpose digital computer is presented. The basic approach involves decomposing the structure of a classical finite state machine into a logic unit, an interconnection array, and a latching unit. The logic unit is implemented with an optical NOR gate array. The NOR gate array involves projecting several binary images on a common surface, inverting, and thresholding the intensity of the result. Each pixel thus functions as a NOR gate. This NOR gate array is sufficient to establish a complete logical set. The interconnect array can be implemented with a hologram or conventional optics. Each pixel (x,y) is imaged to pixel (x,y-1), pixel (x+1,y-1), and pixel (x-2,y-1) on each cycle. This interconnection pattern is sufficient to establish a complete non-planar connective set. The latching unit serves as a memory. A delay is used as a memory in much the same manner as it was in earlier computers (mercury delay lines, CRT storage, etc.). The processor is programmed by customizing the interconnect pattern. This can be accomplished by tying the input of various optical NOR gates "high" with a customizing binary input image. This design illustrates how the parallelism of optics can be used to overcome the Von Neumann bottleneck which throttles the throughput of current computers.

A liquid crystal device that performs a two-dimensional intensity-to-spatial frequency conversion has been developed for use as an optical transducer. When such a device is used as the input transducer in an optical filtering arrangement, image intensity levels can be easily manipulated via Fourier plane filters. This variable input intensity-output intensity transfer function has numerous potential applications in image processing. Implementation of a variable level slice operation is discussed and experimental results presented. The VGM LCLV device is also particularly well adapted to performing binary logic operations on 2-D images. This application is discussed, and results demonstrating the implementation of common logic operations are presented.

The concept of computation modules for an optical residue computer is introduced. The application of the computation modules in mathematical computation, coding and decod-ing is demonstrated. An optimization technique for the module circuit design is also presented. Very high throughput rate is achieved with the use of parallel computation structures and the pipelining of sequential operations.

Methods for improving the accuracy of optical processors are considered. The improvement in accuracy is achieved by increasing the space-bandwidth product requirement of the system. Since convolution is the operation most easily and efficiently implemented by optical systems, we concentrate on systems that achieve convolution with increased accuracy.

Optical fibers and dynamic reflective elements are used to modulate a vector array of optical signals by a matrix of weighting factors, thereby producing vector x matrix multiplication.

This paper deals with an optical arithmetic unit called Numerical Optical Data Processor (NODP), capable of performing addition, subtraction, multiplication and division. The NODP combines the residue arithmetic representation with a spatial light modulator operating as an optically controlled birefringence mirror. The quantity of interest is the relative phase delay between polarization components of light along the fast and slow axes of the controlled birefringence mirror. This phase delay is addititive during several reflections (the basic addition operation) and is detected in terms of the intensity outputting through an analyzer. In the past we have been troubled by slow response times of the LCLV. That has been solved by chopping the input light between two levels. Several addition operations have been performed and results look encouraging. In this paper the operation of the device and some parts of the theory will be discussed and data showing addition will be presented!

In this paper we propose new methods of optically implementing digital logic gates capable of performing all combinatorial logic operations. These logic gates are implemented on a Hughes liquid crystal light valve operated in the parallel off-state. Several configurations for realization of optical gates such as AND, NOR, XOR, etc. will be considered. Experimental results will be given using the portions of a single liquid crystal light valve.

Planetary geological studies are almost entirely based on the analysis of orbital imagery. In the case of Mars, optical power spectra are providing the photogeologist with an additional aid in his task of classification and characterization of diverse terrains. Statistical pattern recognition techniques using optical power spectral data may be especially valuable in subdividing terrain units with characteristics that are only subtly different and in correlation of isolated patches of similar materials that are widely separated on the planet's surface.

Any slowly varying 2-D space-variant system can, in principle, be represented holo-graphically by spatially sampling the input plane and multiplexing the respective system transfer functions. This paper describes an alternative multiplexing technique in which one samples in the Fourier plane to produce the multiplexed holograms. Various techniques for accessing any particular set of the multiplexed transfer functions are considered. Preliminary experimental results are also presented.

The performance / price ratio of digital data processing is steadily increasing, while the performance / price ratio of optical data processing remains nearly constant, although at a favourable level. Given this trend, one might ask: is there still a future for optical data processing? This question cannot be answered in general, since optical data processing is very competent for some jobs, but clumsy or incapable at other jobs. The category of jobs suitable for optics is characterised by features like: high data rate, large storage requirement, moderate accuracy, repetitive program consisting mainly of linear and quadratic operations. Certain statistical computing problems belong into this category. We will present two examples and analyze their data processing efficiencies. The examples are useful in astro-nomy (speckle interferometry) and in biology (motility studies on bacteria).

Scanning image processing systems have the advantage that the image is in a form suitable for further electronic processing, and a particular class of these is the confocal scanning systems which exhibit superior imaging properties compared with conventional systems. Imaging is coherent thus allowing coherent processing to be performed with twice the spatial-frequency cut-off of a conventional processor. The point spread function is also sharper and the outer rings weak, resulting in a response to a step object without fringes. The important feature of the confocal system is that two lenses are involved in the imaging process. An arrangement in which the beam traverses the object more than once is therefore also discussed. This produces an extremely sharp point spread function and interesting optical properties.

The general properties which an optical system must possess to display amplitude or phase contrast of both differential and non differential form are derived. These results are applied to various forms of scanning image processing systems and it is found that by suitable choice of lens or detector distribution that the desired contrast mode may be obtained. The coherent confocal system is particularly suited to the methods of coherent processing with the additional advantage of possessing twice the spatial frequency bandwidth of conventional systems.

The increasing complexity and variety of image sensors has been the source of interest in the development of data compression for images. Image data has become one of the most active topics of research in digital image processing as a result [1]. The continued evolution of digital circuitry has caused the focus of data compression research to lie in digital implementations. However, there is also a potential for optical computations in image data compression, as was demonstrated in the concepts of interpolated DPCM [2]. The method of DPCM data compression is one of the most thoroughly studied techniques. DPCM achieves data compression by separating the image information into two parts: the low-spatial frequencies and the high-spatial frequencies. Low-spatial frequencies are re-tained by exploiting their predictability; high-spatial frequencies are retained at fewer significant bits, and substantial data compression is achieved. Interpolated DPCM is a mechanism for separating an image into low- and high- spatial frequency components, with a similar amount of data compression being achieved. The computations to achieve the sep-aration can be implemented by simple incoherent optical devices [2].

Coherent light propagating along a multimode optical fibre may be phase modulated by external disturbances including changes in for instance the pressure or temperature of the medium surrounding the fibre. At the output from a multimode fibre, a direct phase to amplitude modulation conversion effect takes place at the detector, and this signal may be used to detect the presence of the environmental change. However, the phase detection process is intrinsically non-linear and is subject to fading effects. This paper concentrates on describing techniques whereby reliable and consistent phase detection techniques may be implemented by optical means and presents the results obtained in the actual use of some of these techniques.

Two applications of a Fabry-Perot interferometer in image processing are reported. In the first, the angular selectivity of the device is exploited in order to filter the angular spectrum of the input. Operation such as tunable low, band or high pass filtering are demonstrated. The second application uses the fringes of a detuned Fabry-Perct as a spatial carrier frequency tunable in both frequency and orientation. An application in density slicing is demonstrated.