The ERIM Thermoplastic Optical Phase Recorder (TOPR) is a real-time, two-dimensional light modulator. The modulator employs electron beam writing on a deformable and reusable thermoplastic film. The varying thickness film causes phase modulation of a light beam in an optical processor. The data written into the plastic can be stored for periods from less than 0.5 sec to many months depending on the temperature at which the plastic is maintained. The time bandwidth product of the device is greater than 106 and its optical quality is suitable for optical data processing. Because of these features the TOPR, combined with an optical data processor, can offer advantages in size, cost, complexity, and processing speed relative to digital computers. Alternatively, the TOPR can be used in a hybrid optical-digital computer to provide preprocessing of data or a "quick-look" capability for digital data processing decisions. This paper describes characteristics of the TOPR and an application of the TOPR to processing of imagery from data recorded in a unique polar format. Introduction The purpose of this paper is to describe a real-time spatial light modulator and its use in a coherent optical computer in a practical application; synthetic aperture radar (SAR) data processing. In this modulator a scanning, intensity-modulated electron beam is used to deform a thin thermoplastic film. A large area light beam is passed through the film (or reflected from it) and the deformation image is impressed onto the light beam as a two-dimensionql phase modulation. This modulator was developed specifically for optical data processing', thus, it has good optical quality, large space-bandwidth product, near real-time response, and a reusable modulation medium. In the following we will discuss the modulator design, its specifications, and its performance in the SAR data processing application.
Real-time data storage and processing using optical techniques have been considered in recent years. Of particular interest are photosensitive electrooptic crystals which permit volume storage in the form of phase holograms, by means of a charge transfer process. A survey of the state of the art of such holographic memories is presented. The physical mechanism responsible for the formation of phase holograms in such crystals is discussed. Attention is focused on various aspects of materials characterization, development and utilization. Experimental reversible holographic read-write memory systems with fast random access and high storage capacity employing this new class of photosensitive materials have already been demonstrated.
In recent years it has become possible to read out electronic image sensors by means of deformable membrane mirror arrays1'2"'"5. Similar techniques have also been explored for use with scanned electron beams6'7. The principal motivation for this work has been "real-time" light modulators for the input plane of coherent optical processors.
We report here the feasibility of coupling an image intensifier tube directly with the ac liquid crystal light valve, which is briefly described and characterized in the paper, to form a compact, real-time, incoherent-to-coherent image converter. This hybrid device is capable of sensing image intensity levels of less than 10-6 In addition, the same combination can be used as an image amplifier with an intensity gain of up to 10. In the experiments performed we mated an 18 mm aperture micro-channel plate inten-sifier image tube, with a fiber optic faceplate output, with a 50 mm aperture hybrid field effect liquid crystal that has a fiber optic faceplate input. With an incoherent input image intensity of 3 x 10 ft-candle we observed a limiting resolution of 16 1P/mm in the output coherent image. This resolution limit is determined primarily by the image intensifier tube. MTF data for the hybrid device and the components are also presented, along with a photograph of gray scale image capability. The use of other types of image intensifier tubes, such as those having image storage capability, with the light valve is discussed and performance ranges given. Next we summarize the possible uses of the hybrid device to increase the usefulness of real-time optical data processing. Finally, we describe the application of this device to wavelength conversion and image amplification to provide the large screen projection of dynamic ac plasma panel display imagery.
Information storage, including image storage with gray scale, has been previously reported for PLZT ceramic ferroelectric-photoconductor (FE-PC) devices in which the infomation is stored as spatial variations either of birefringence or of light scattering centers. The FE-PC devices also store information in the form of variable surface deformation. We have discovered recently that nonvolatile, gray-scale images, superior in quality to those achieved with FE-PC devices, can be stored in rhombohedral-phase PLZT ceramics (without photoconductive films) by simultaneously exposing the image on a surface of a ceramic plate using near-UV light and switching the ferroelectric polarization through a portion of the hysteresis loop. This paper reviews some important characteristics of the FE-PC devices and describes the new photoferroelectric image storage mechanisms and devices.
The PROM is a solid state, rapidly recyclable, image storage device having a number of applications in image and signal processing. Some of its important characteristics include 10th-wave optical surface quality, 100 1p/mm three-bar resolution, 10-ergs/cm2 light sensitivity, and image plane contrast of 104:1. One of the unique features of the PROM is that the bias level of stored patterns can be adjusted through application of an external voltage, resulting in image contrast inversion or enhancement. This same operation (baseline subtraction) is used to null the zero order in an optical Fourier transform, achieving a Fourier plane signal-to-noise ratio approaching 106:1. This paper will report on the current status of this device and a number of applications for which it has been tested in several areas of image and signal processing. Results will be shown for coherent optical processing by computer controlled Fourier plane filtering and real time image correlation, and sig-nal processing systems will be described which couple the PROM with an acousto-optic raster recorder to perform spectrum analysis and correlation on radio frequency signals.
Ruticons are optically addressed light valves that are proving increasingly useful for image storage and optical processing applications. A wide range of image storage times and sensitivities are obtainable, depending upon the desired application. High quality Ruticons have been constructed that are well suited for both noncoherent and coherent optical processing operations.
Coherent optical feedback systems are Fabry-Perot interferometers modified to perform optical information processing. Two new systems based on plane parallel and confocal Fabry-Perot interferometers are introduced. The plane parallel system can be used for contrast control, intensity level selection, and image thresholding. The confocal system can be used for image restoration and solving partial differential equations. These devices are simpler and less expensive than previous systems. Experimental results are presented to demonstrate their potential for optical information processing.
By their basic two-dimensional nature, coherent optical systems provide an additional degree of freedom for the processing of one-dimensional signals (e.g., speech, radar, and general communication signals). This second degree of freedom has been used in the past to increase substantially the information throughput of optical processors for one-dimensional signals. We describe here how it can be exploited to increase the kinds of operations that can be performed optically, concentrating on signal processing operations that are frequency-variant. One example considered is the frequency-variant spectral analysis of signal waveforms, a specific case being a log-frequency constant proportional bandwidth spectrum analyzer. Also discussed are applications of the basic concepts, combined with optical heterodyne techniques, to linear and nonlinear bandwidth compression. Analytical and experimental results are presented.
Space variant image processing using combined geometrical transformations and optical transforms is discussed, whereby the flexibility of an optical processor is enhanced. Use of computer generated hologram masks to effect the geometric transforms is emphasized.
A method for real-time matrix multiplication is presented. This paper describes the geometrical interpretation of the mathematical manipulations between the two matrices. Three coherent optical astigmatic systems are developed based on the analysis. Each system is essentially composed of two subsystems that are con-nected in series. The first one performs multiplications between the corresponding elements of the matrices coded in the amplitude transmittance of the transparencies. The results are received by the second subsystem that performs the necessary summation operations to give the calculated rise to each element in the final result, the product of the two matrices. In these processes, no preparation of a hologram or intermediate memory is required. The operations are done in parallel. The multiplication between an N x N matrix and an N x 1 vector is presented in detail. The possibility of multiplication between N x N and N x N matrices is also discussed.
The Goddard Space Flight Center is presently developing hardware technology for building ultra high speed computers for processing two dimensional data. This technology employs massive parallelism, doing thousands of operations simultaneously, to obtain this speed. The operations are performed over a binary image or matrix which is the basic data entity. If the image operations are performed at the same speed that a standard computer operates on a single binary digit the image processing speed is potentially increased by a factor equal to the number of elements in the binary image. Computers that operate on binary images have come to be called tse computers. The word "tse" is Chinese for their writing characters which are actually stylized images.
From Gabor's concept of information and Shannon's discrete channel, the capacity of an optical spatial channel is calculated. We have illustrated two basic spatially image encoding processes; namely, the encoding over a set of uniform resolution cells, and over a set of nonuniform resolution cells. One of the important results is that the most efficient image encoding is the one that provides the most probable encoded irradiance distribution over the spatial channel.
Optical analog systems which are capable of doing nonlinear processing have potential application in several areas including equidensitometry and analog-to-digital conversion. One approach to nonlinear optical processing is derived from the idea of halftone printing. This approach to nonlinear processing requires custommade halftone screens. It is important to be able to produce such screens accurately and reproducibly. This paper describes a technique of making a binary-coded halftone screen. The desired continuous-tone screen is obtained as an aerial image by spatial filtering the binary mask. Techniques of microlithography have been advanced to such a state that it is possible to produce binary masks which cannot be adequately imaged by lenses. This paper discusses a method of circumventing this limitation. This is done by a means of self-imaging which incorporates the equivalent of low-pass spatial filtering.
Pseudo-coloring of a photograph in a coherent optical processing system can be achieved through the utiliza-tion of the newly advanced technique of halftone screens. The basic principle of the pseudo-coloring method is to mix the high diffraction order outputs of the halftone photograph generated by lasers of different wave-lengths in a coherent optical system. A discussion of the theory, the method, and its potential applications will be given.
A generalization of the halftone screen process for achieving nonlinear functions in coherent optical systems is given. Using non-monotonic periodic halftone screens, functions with an arbitrary number of slope changes can be obtained in the first diffraction order. A detailed synthesis algorithm for specifying the screen shape is described, and some preliminary experimental results on real-time nonlinear processing are presented.
A scheme for realizing certain geometrical distortions in a coherent optical system has been proposed and demonstrated by Bryngdahl. This technique is described as being applicable when the object functions exhibit "relatively slow spatial variations." In this paper the notion of slowly varying object functions is examined in terms of the spatial bandwidth of these functions, thus providing a means for examining the merit of such a distortion technique. It is shown that the technique may be expected to work well at low spatial frequencies, worsening with quadratic spatial frequency dependence. The spatial frequencies at which the distortion technique fails may be readily estimated. In general, these limiting spatial frequencies depend on wavelength, separation between input and output planes, posi-tion in the output or input plane, and the particular distortion one wishes to realize.
Optical cross-correlation to determine relative signal displacements and degree of similarity between two signals is commonly implemented by matched filter techniques using absorption transparencies as inputs. The problems associated with absorption inputs include law correlationsignal intensities due to the light absorbing nature of the input and law signal-to-noise ratios. Without considerable preprocessing, positive correlation peak detection is not always readily achievable. These limitations are largely overcome by complex exponentiation of the inputs. For the optical analog this means phasing the input transparencies by a bleaching process to yield phase transparencies. The cross-correlation function of these complex exponentiated inputs has two striking properties. One, the correlation-signal approaches a delta-function. Two, the correlation-signal is not affected by a difference in bias levels (average densities) of the two inputs inasmuch as only differential phase differences are used for detecting correlation. This means constant phase shifts will not contaminate the correlation signal. Hence, extensive data preprocessing is not required. One- and two-dimen-sional digital simulation experiments were carried out to demonstrate these properties. Simulated density functions were defined by computer generated random numbers. Random noise distortions were added to study their impact on the correlation-signal shape and intensity. In order to have a standard for comparison, the commonly used (statistical) correlation coefficient was computed along with the correlation of the complex exponentiated inputs. The results indicate that complex exponentiation provides a means to obtain extremely reliable correlation peak positions having very high peak intensity and very high signal-to-noise ratio (SNR). Since the correlation function is a narrow well defined spike, a threshold detector can be employed for signal detection.