In recent years, several new techniques have been proposed for the synthesis and implementation of correlation filters which are philosophically quite different from earlier methods. This paper provides an updated overview of the design to correlation filters and summarizes these newer techniques and their advantages.

We previously reported computer code designed to generate optical correlator filters by putting our optimization algorithm into practice. This code is MEDOF: Minimum Euclidean Distance Optimal Filter. MEDOF's version 1.0 demonstrated filters which maximize correlation peak intensity (INT). We also gave some discussion to the quadratic ratio metrics of signal to noise ratio (SNR) and peak to correlation energy ratio (PCE). These advanced metrics are now functional in the second version of MEDOF, which is publicly available through NASA's software distribution channel, COSMIC. This paper focuses on some of the practical issues faces when using MEDOF to generate filters. The underlying mathematical principles of MEDOF are discussed in passing while providing references to sources of more theoretical detail.

We consider detection (location of all likely regions of interest (ROIs) in a scene where objects may be) for multiple classes of objects in 3D distortions with contrast differences and severe clutter present. Two different algorithms using Gabor basis function (GBF) filters and morphological wavelet transform (MWT) filtering are considered. New final algorithm parameters are noted. We detail: the morphological portion of the MWT algorithm, our new fusion method to combine the morphological and Gabor wavelet clutter map portions of the MWT algorithm, and the MWT threshold selection technique. Initial results using a new peak sorting scoring method and new fusion scores for multiple algorithms to reduce false alarms are noted.

The minimum average correlation energy filter (MACE) filter has been shown to have superior performance for rejecting out of class inputs in pattern recognition applications. The MACE filter exhibits a sharp correlation peak at a specified location in the output plane and low correlation energy elsewhere. It has also been shown that the MACE filter suffers from poor generalization. Increasing the number of exemplars used to compute the filter coefficients can improve the generalization, but the number of exemplars is restricted by the stability of the computation. We show a simple extension of the MACE filter to nonlinear processing techniques (i.e. nonlinear associative memories) which exhibits improved generalization and discrimination performance. The operating parameters of the proposed extension are difficult to compute analytically and adaptive learning methods are needed. Since the output of the MACE filter is optimized over the output plane any nonlinear extension of the MACE filter should encompass the output plane as well. In general this leads to exhaustive training over the entire output plane over all training exemplars. We present an efficient method for computing the parameters of the nonlinear extension which greatly reduces the training iterations required. Experimental results with 35 GHz inverse synthetic aperture radar (ISAR) data are also shown.

We consider a new detection application for distortion-invariant filters. Several new advances to the MINACE filter are considered. These include: improved object modeling (this provides scene correlation plane peaks near the specified values), improved clutter modeling (no false class training is used), a new trade-off parameter c definition (with the energy of each spectra normalized to one), use of a smaller filter size and a prime factor FFT (this reduces noise effects), zero-mean filters (these allow detection of hot and cold contrast objects), etc. We advance a new peak variance degree of freedom distortion-invariance filter (PVDDF) with many attractive new properties. These advantages include: insurance that the correlation peak values for all distorted objects are close to a given value without requiring that each object have a given correlation peak value (in practice one does not want all distorted object inputs to give the same exact correlation peak value), use of more training images N_T without an associated drop in the required threshold and thus better object modeling, and better object function energy E minimization (E reduces as N_T increase compared to other filters where E increase as N_T increases). This new filter thus achieves both better correlation peak values and better energy minimization (prior filters cannot achieve both goals) by using degrees of freedom.

We present an approach to estimating the distortion parameters of distorted target objects. We first detect the presence and position of the targets with a bank of distortion-invariant correlation filters. The filter set we use, known as hybrid composite filters, yields complex responses at the target locations. The phase angles at the target locations are linearly related to the distortion parameter. In the case of in-plane or out-of-plane distortion, the distortion parameter is the rotation angle of the object. We assume a finite number of values for the distortion parameter so an M-ary classification algorithm is used to determine the input distortion parameter. The classification algorithm is a Maximum Likelihood M-ary classification algorithm, more commonly known as 'largest of' filtering, which determines the most likely distortion parameter value. A set of complex signatures is formed with weighted linear combinations of the different filter target responses. The signature with the highest correlation with the set of known signatures indicates the associated distortion parameter. In practice, we do not use all the hybrid filters and the ones we do use are optimally selected and combined for detection and discrimination. Since we know the target location, we can use a different set of filters, optimized for angle estimation, and implement them with an inner product operation. Inner product filter responses are complex so estimation ambiguities may arise. We present a selection criterion, optimal for signal-to-noise ratio. We combine knowledge of the ambiguities with this optimal selection criterion to prevent ambiguities and achieve minimal estimation error. Numerical simulations are presented to demonstrate this new approach to parameter estimation.

The phase-only filter is of great interest for optical pattern recognition applications because of its sensitivity and because it can easily be programmed onto spatial light modulators. In evaluating the suitability of this filter for a correlator, it is useful to understand the impulse response function that is actually being correlated with the input function. In this research, we derive an analytic expression for the impales response function for the phase-only filter corresponding to a one-dimensional rectangle function. We show that this impulse response function represents a bipolar edge-enhanced version of the input function. This can be easily generalized to a two-dimensional box function. Experimental results verify the analysis.

A SDF (Synthetic Discriminant Function) could not cover a large distortion range to ensure a good correlation S/N with training images. There exists a limitation for the training image number of a SDF. To extend the distortion range of SDF hardware implementation system, a new spatial multichannel technique was proposed. A multiplex SDF real-time 3D distortion-invariant spatial filtering was realized using holographic technique by means of a special phase grating, a high correlation S/N, very good abilities of intraclass recognition and clutter rejection were got, multiobject tracking and shift invariant abilities were kept. The 3D object OPR (Optical Pattern Recognition) distortion-invariant range was extended greatly. This technique could be applied to practical machine vision purpose.

The Weapons Sciences Directorate has dedicated many years of research to the development of optical correlators for applications in automatic target tracking and recognition. A great deal of effort has been directed toward the implementation of compact optical systems, optimal filter technologies, and high frame rate modulators. This effort follows general trends in the industry to improve the quality of the correlation plane. However, relatively little attention has been directed toward automating the analysis of images obtained at the correlator's output. The existence of a reliable post processing scheme is a prerequisite to the construction of a fully automatic correlation system. The design, construction and programming of a hardware post processor has recently been completed, and preliminary tests of the device have been successful.

For realizing machine vision using optical coherent information processing technique, distance and direction of moving target must first be detected. In our digital-optical hybrid correlation filtering system, a new method of detection distance and direction of moving target was proposed. In the paper, we studied the algorithm of detecting distance and direction of moving target by means of the method and gave the experiment results. Without question, it would have practical application in the study of machine vision.

The general approach to the problems of recognition and classification of objects based on the analysis of its anisotropy properties is discussed. Anisotropy properties of objects are presented by seven independent parameters having clear physical sense: relative value p and orientation (theta) of linear amplitude dichroism; relative value r of circular dichroism; relative value (delta) and orientation (alpha) of linear birefringence; value (phi) of circular birefringence and value of depolarization P. Analysis of anisotropy properties and obtainment of concrete values of above parameters are based on the theorem of decomposition claiming that the nondepolarizing Mueller matrix can always be presented on basis of the matrices of the phase and amplitude anisotropy. Developed in details classification system is presented. Taking into account the facts of presence or absence of any sort of anisotropy and of depolarization any object can be referred to the one of the 32 classes. If information on isotropic changes of amplitude and phase of radiation interacting with an object is interest than the analysis will be carried out in extended basis of anisotropy. Extended basis of anisotropy is the full equal model of real medium.

Principles for analog and digital optical processing based on the two- pulse photon echo (PE) phenomenon are considered. Various architecture designs of optical pattern recognition systems and optical processors for analog and digital processing are discussed. Also various schemes of digital processors for vector-matrix multiplication with the pixel structure and of holographic type are suggested. These realizations depend on using of digital temporal or spatial mixed binary multiplication based on the PE.

An all-optical wavelet processor for target detection has been developed and demonstrated at JPL. An innovative 2D Shape-discriminant wavelet filter, encoded with a ternary-valued scheme, was developed and implemented in a Liquid Crystal Television Spatial Light Modulator (LCTV SLM) to enable real-time updatable processing. Applications ranging from flight object discrimination, mine detection, and cancer cell detection have been demonstrated.

In this paper, a multichannel optical wavelet processor and a matching pursuit processor capable of enhancing the detection of cluttered targets are presented. Wavelet functions have zero-mean and are virtually band-pass filters. In many cases, targets and clutter are separable in the spatial spectral domain. Therefore, by selecting wavelet functions that represent features of targets but are insensitive to that of clutter, targets can be extracted from the input scene while clutter is suppressed. Due to dyadic sampling, a multichannel optical wavelet processor with a limited number of channels can detect regions of interest for different targets. With matching pursuit decomposition, features of targets are extracted and represented in a few wavelets known as coherent structure; whereas clutter and noise are diluted across the dictionary. Clutter and noise can then be effectively removed from the signal by a simple thresholding operation. A time-frequency energy distribution can be derived from matching pursuit decomposition, which contains no interference terms and thus clearly characterized the input signal in the time-frequency plane. Optical architectures of these processors are described. Simulated and experimental results are provided.

Wavelet based Joint transform correlator (JTC) for rotation invariant automatic target detection is proposed in this paper. Wavelet features of a set of rotationally distorted training images are first extracted at different levels of resolution. A simple composite filter formulation is then employed to construct the reference image of the JTC. Simulation results are presented showing the improved rotation-invariant detection performance of the proposed technique.

In conventional approaches to the design of binary phase-only synthetic discriminant function filters the filters are trained on discrete Fourier transforms (DFTs) of the training image, even though DFTs are in fact downsampled approximations of the continuous optical Fourier transform (FT) that is generated in an optical correlator. The justification for this is that one can completely reconstruct the training image from its DFT. In this work however, we show, by use of a realistic correlator simulation, that filters designed by the conventional approach do not give the correct results when implemented in a real correlator. It is shown that multiresolution wavelet analysis approximations (MWA) of the training image FTs should be used for correct filter design. Furthermore, we show that it is possible to design filters whose resolution is reduced with respect to the conventional case, but whose performance is comparable, using the MWA approach. Finally, the performance of the filters trained on MWA approximations was found to be vastly superior in nearly all cases to the equivalent filters trained on conventional downsampled approximations.

We used the wavelet representation of an image for multiresolution processing. As the resolution of an object was decreased we maintained the signal-to-noise ratio (SNR) at the highest resolution by changing the support region of a binary phase-only filter (BPOF). In this way we were able to decrease the resolution by a factor of four while maintaining the SNR at the full resolution for our test image. Although the SNR remained constant as the resolution decreased, the peak-to- correlation energy decreased. The processing of an object at a lower resolution is significant because it corresponds to processing the same object in a larger image after downsampling. Due to the limited number of pixels in currently available SLMs, multiresolution processing could extend there usefulness.

We introduce the circular harmonic wavelet matched filter, that uses the wavelet transform in improve the performance of the rotation invariant circular harmonic filters in terms of the peak sharpness and the robustness to noise. In the optical implementation, the wavelet bandpass filter guarantees the full range rotation invariance of the filter by eliminating the circular subsampling error.

The authors' previous work (SPIE Vol. 2237) on scale-, rotation- and shift-invariant wavelet transform is extended to accommodate multiple objects in the scene and a nonuniform background. After background elimination and segmentation, a set of windows each containing a single object are analyzed based on an invariant wavelet feature extraction algorithm and neural network-based classifier.

The theoretical development of purely real composite filters for optical correlators is proposed. These filters can be useful if the spatial light modulators (SLM) offer an 'amplitude only' mode, and provide an alternative to the fully complex ideal filters which cannot be implemented in general due to the limitations of the modulation capabilities of most existing SLMs. We analytically show that the performance of the proposed real filters is at most 3dB below that of their ideal counterpart. Even if their applications remains limited due to properties of real-world SLMs and the advent of mapping techniques to fit complex filters to device characteristic curves, the study of purely real filters may offer insights to establishing lower bounds on the performance loss for other limited modulation filters.

For optical implementation of correlation filters, we must design filters that can be implemented on available spatial light modulators (SLMs). These SLMs are usually characterized by the fact that their amplitude and phase responses are cross-coupled and thus cannot be independently controlled. Recently, we have introduced optimal trade-off filters that can be implemented on such cross-coupled devices. In this paper we present simulation results using these filters.

Design of optical pattern recognition filters taking into account the nonideal characteristics of the spatial light modulators on which the filters are implemented is an important research problem. In this paper, an iterative method is developed for the design of SLM constrained minimum average correlation energy (MACE) filters. The algorithm uses a relaxation algorithm in conjunction with Juday's minimum euclidean distance (MED) mapping technique in an iterative manner. The performance of the filter designed using this method was evaluated using computer simulations and the results are compared with a constrained MACE filter designed using a software based on a simulated annealing technique. The new software requires much less computer time than the simulated annealing based software providing comparable response. The time taken by the new algorithm is more than that for the MED mapped design; but, the new algorithm provides less deviation from the specified response for training images than the MED mapped design.

Composite filters offer one solution to the problem of scanning numerous filters to find a single target at any aspect, orientation, or configuration. Many different techniques for generating composite filters exist, each with their own advantages and disadvantages. Liquid crystal televisions are an increasingly popular choice for spatial light modulators and are characterized by a coupled phase and amplitude response. The coupled response can cause a problem if not accounted for, however, it can also be used to more closely approximate the desired filter response. This paper will present the results of using optimally encoded composite filters on a LCTV for optical pattern recognition applications.

Liquid crystal televisions are popular low-cost spatial light modulators. One LCTV of interest is found in the InFocus TVT-6000 television projector. A wavefront splitting interferometer has been constructed and analyzed for measuring the complex characteristics of these modulators, including phase and amplitude coupling. The results of this evaluation using the TVT-6000 projector drive electronics have been presented in a previous work. This work will present results of the complex characterizations of these modulators using custom drive electronics.

The paper presents problems related to automated recognition of printed music notation. Music notation recognition is a challenging problem in both fields: pattern recognition and knowledge representation. Music notation symbols, though well characterized by their features, are arranged in elaborated way in real music notation, which makes recognition task very difficult and still open for new ideas. On the other hand, the aim of the system, i.e. application of acquired printed music into further processing requires special representation of music data. Due to complexity of music nature and music notation, music representation is one of the key issue in music notation recognition and music processing. The problems of pattern recognition and knowledge representation in context or music processing are discussed in this paper. MIDISCAN, the computer system for music notation recognition and music processing, is presented.

We have recently shown that the processing of compressed and encrypted imagery can achieve computational speedup and data security by processing fewer data, which are encoded in an obscure format [5- 9]. Our previous work in compressive processing produced numerous image processing algorithms that yielded computational speedups on the order of the compression ratio. In Part 1 of this series [1], we discuss the theoretical basis for pattern recognition over compressed imagery. In this paper, we present theory in support of optical or electro-optical implementations of convolution or correlation operations over block-, transform-, and runlength-encoded imagery, as well as data encoded by vector quantization (VQ). Unlike our previous work in this area, we do not derive operations that return a compressed result. Instead, our algorithms produce a map of correlation coefficients in the image domain, using a compressed image as input. Several of our architectures could, in principle, perform in time that is at least proportional to the compression ratio. Theory is expressed in terms of image algebra, an emerging branch of mathematics that unifies linear and nonlinear mathematics in the image domain. Image algebra has been implemented on a variety of workstations and parallel processors, as well as electro-optical processors. Thus our algorithms are feasible as well as portable Analyses emphasize computational complexity and information loss.

A new binary joint transform correlator (BJTC) algorithm has been implemented which compares a target fingerprint with six reference fingerprints. The correlator uses the frame subtraction technique to achieve very dense packing of the images in this input plane. Six 128 X 64 reference prints and a target print are displayed simultaneously on a 256 X 256 input SLM. A peak-to-noise ratio (PNR) of 12.7 dB has been demonstrated experimentally, and a potential for a PNR as high as 41 dB is shown using simulation. This is compared to a PNR of 6.8 dB achieved without using frame subtraction. This demonstrates that through use of frame subtraction the input separation constraints for the BJTC are eliminated and performance is improved.

A novel BPEJTC-based method of extracting a moving target's signature residing in a background input is proposed. The efficiency of this method depends to a great extent on the degree of a JTC performance. A BPEJTC is shown to have a improved peak-to-sidelobe ratio than that of a conventional JTC. An optoelectronic system implementing this method is realized. In this system, time-consuming computations, like FFT are conducted optically while arithmetic operations are executed digitally such that the proposed hybrid system can operate in real-time. Some experimental results obtained with real input images indicate that the proposed system can effectively work even in a nonstationary background.

In this paper we consider linear correlation filters for image pattern recognition, with particular application to Synthetic Aperture Radar (SAR). We investigate the statistical properties of several popular Synthetic Discriminate Function (SDF) based linear correlation filters, including SDF, MVSDF, and MACE filters. We compare these statistical properties both qualitatively and analytically for SAR applications. We also develop modifications to these SDF-type filters which have particular utility for Synthetic Aperture Radar (SAR) image classification. We compare the performance of the modified filters to the standard filters using X-patch generated SAR images with both white and colored noise. We also investigate effects of performance degradation caused by mis-estimated noise statistics, and the effects of image normalization on the target detection rates.

Most distortion-invariant optical pattern recognition techniques rely on correlation which inherently achieves translation-invariance. We introduce a new formulation for image recognition where only 'vector inner product' (assuming 2D images are lexicographically converted to vectors) operations are used to achieve distortion-invariant pattern recognition. Our formulation expands the linear phase coefficient composite filter family, developed by Hassebrook et.al., into a set of translation- and distortion-invariant vector inner product operators. The set of vector inner products are optimal in numerical efficiency because they represent a Karhunen Loeve expansion. Translation- invariance is achieved by embedding 2D translation into the training set as two additional distortion parameters. The magnitude of the vector inner product results in a response insensitive to translation and distortion, where as the phase response varies, but is discarded. For large images containing many objects this method can be applied by tiling the vector inner product operators to the test image size. Examples of our approaches, distortion-invariant detection/discrimination capabilities, numerical efficiency, and tradeoffs between conventional correlation are presented.

We use color for visual presentation of certain effects that are complex in nature but have previously been shown using only amplitude display. In particular we show (1) the field that results from complex encoding of an input object, and (2) the departure from ideal of the object that is actually correlated with the input object. The departures are caused by limitations of filter SLMs and by the algorithms that compute filters for them.

The application of a multifunctional extended optoelectronic correlator (MEOC) system in the field of pattern recognition is presented. The MEOC device is based on the extended optical correlator (EOC) architecture in conjunction with a digital image processing system. The EOC system is a three-lens coherent correlator with three separate planes usable for in- line spatial filtering of signals. Combining amplitude spatial filtering in both the frequency and image planes with complex filtering in the matched spatial filter (MSF) plane, the MEOC system was used for performing various complex procedures in the pattern recognition area. Furthermore, the MEOC device was advanced by inserting an analogue coherent optical averaging (ACOA) setup. Subjects of interest were cuneiform inscriptions on an original Babylonian cuneiform tablet. The investigations using the MEOC system were carried out in the following steps: feature extraction, average pattern mask production, average matched spatial filter production, and finally the correlation experiment. The results show that classical MSFs of averaged objects combine a low in-class sensitivity with a high discrimination ability for out-of-class objects, if suitable preprocessing steps have preceded.

Automatic target recognition has a wide range of applications spanning many different imagery formats. These input formats include, but are not limited to video, infrared imagery, and synthetic aperture radar imagery. Optical correlators, while not the total ATR solution, have the potential to aid in this complex task. One area of recent interest has been near and far field effects in synthetic aperture radar imagery. An optical correlator has been used to qualitatively assess the differences in imagery obtained for both the near and far field cases. Reference filters were made for each case and compared to imagery from the opposite case. The results of these experiments will be presented.

This work presents an approach to the detection of weaving defects by optical correlation. The principle of the method is to realize a global comparison by optical inter-correlation between the flawless reference fabric and the fabric which one wants to test for defects. It requires the recording on a holographic plate of the diffraction spectrum of the flawless fabric, according to the principle of analog Fourier holograms. The hologram thus recorded represents a memory for a type of tissue without defects. It constitutes the reference element constantly compared to the spectrum of the fabric which is searched for defects. In the absence of defect the autocorrelation function is obtained. The amplitude and shape of the correlation function change when a defect appears.

The problem of video and infrared sensors calibration for mutual inspections in accordance with the Open Sky Agreement is considered. The fact that the validity increase and accuracy of calibrating sensors can be achieved by means of automation of recognizing images of test-pattern is shown. For this purpose Hough transform is used. The approach in the form of algorithm is realized. Performance of the algorithm on aircraft imagery shows its good convergence with visual method. Realization of the developed approach by using digital and discrete-analog processors is discussed.

Mathematical model of optoelectronic image array processor for SAR system is described with consideration for disturbing factors which are present in real optoelectronic system, such as noise and errors of light sources and photodetectors, ADC nonlinearity and point spread function of the optical system. Analytic formulae relating processor accuracy to the value of error probability and accuracy parameters of its processor elements as well as optical system quality are delivered. The obtained theoretical results are closely approximated with modern experimental data.

The states of polarization of the light after reflection and transmission through the sandwich structure of the refractive layers are analyzed. Calculated by Wolter formula, the complex transmission coefficient of multilayer system indicates that the transmitted light, having linear polarization at incidence, is elliptically polarized. The ellipticity and the azimuth of the transmitted light is connected with K and L coefficients which represents the real and imaginary parts of the inverse transmission coefficients values. These coefficients can be calculated for the parallel and normal field components, from the found recursive formulas and for any sandwich consisting of m boundary planes between (m-l) layers. From the calculated expressions different number of layer plates, is shown that the transmitted light is elliptically polarized, except at normal incidence and when the incident light azimuth is 0 or 90 degree. Under these conditions the transmitted light preserves its linear polarization and the azimuth of the vector field vibrations. The reflection case is analyzed by using Wolter formula as well.

A photothermoplastic (PTP) carrier capable of selective data rerecording and diffraction efficiency correction is considered. A multichannel correlator for pattern recognition using the neural training algorithm is made on its basis.

Optical associative memories have been demonstrated previously that store an accumulation of outer products. Each outer product represents an associated vector pair. In this paper we use a joint transform correlator with a liquid crystal (LCLV) to form the outer products which are then accumulated by mapping with Argon light onto electron trapping material (ETM). Argon light causes electrons to be trapped at the ETM for later recovery. Further we show for the first time that we can remove the outer product for any one of the associated pairs from the accumulated stored matrix of accumulated outer products. This is accomplished by placing the pair to be removed at the input and using IR instead of Argon light to map the outer product onto the ETM. IR light removes stored electrons in the ETM. Experimental results show the practicality of this approach.

A method for optical pattern recognition which is based on the human visual system and is suitable for hardware implementation is presented. The system is composed of two stages. The first stage detects local features such as line orientation, linestops, corners, and intersections to create a feature map, which represents the number of these features and hence is invariant to position, size, and slight deformation of an input pattern. The next stage is a multilayered neural network that classifies an input pattern to one of predefined categories using the feature map. We have found a method of detecting these features in analog hardware which would considerably speed up the process of pattern recognition. The decomposition of an input pattern into lines with different orientations is done by an array of two-dimensional orientation sensors. We have built an orientation sensor which is invariant to the position, size, and contrast of an input pattern. The generation of the feature map is currently being done in software which receives its inputs from the line orientation sensor. Linestops, corners and intersections are detected after a series of convolution and thresholding operations for each orientation. The convolution operation can be mapped into hardware using a resistive grid technique. The simulation with an example of character recognition showed that the proper selection of convolution kernels and thresholds can detect local features described above and demonstrated the feasibility of a full hardware implementation of a feature detector.

An analog 128 X 128 spatial light modulator (SLM) has been designed and constructed using liquid crystal on silicon technology. This device is loaded with eight-bit grey-level data in 100 microsecond(s) . Its pixel pitch is 40 micrometers giving an array size of 5.12 X 5.12 mm. Low-voltage ferroelectric liquid crystals are used for the electro-optic modulator. These analog materials have 50 microsecond(s) to 100 microsecond(s) switching times, implying a frame rate of approximately 5 kHz. This paper presents results of the analog SLM and discusses modulation enhancements for improving correlator performance.

Measurements have been made of the phase modulation and polarization rotation of some commercially available rewriteable magneto-optic compact disks (RMOCDs). The disks were then used in an optical processor (joint transform correlator) as spatial light modulators.

An architecture for combining two twisted nematic liquid crystal arrays (LCTV's) to independently control amplitude and phase modulation of light is examined experimentally. The arrays are driven with the complex information of the inverse Fourier transform of amplitude-only images. Optical Fourier transformation is used to restore the amplitude images. Methods for experimental determination of amplitude and phase properties of LCTV's are also detailed.

We obtain electrically switchable optical complex amplitude modulations with two commercial twisted nematic liquid crystal televisions (LCTV's) which are in the phase/amplitude coupled modulation modes. The condition for the full complex modulation is that one of the coupled mode LCTV's provides the fill range of 2(pi) phase modulation. A look-up table encoding method is proposed that permits to compensate and correct the coupling and nonlinearity in the modulations of the individual LCTV's. A simple wedged share plate interferometer is proposed to measure the phase modulation. Based on the theory of the twisted nematic LCTV, experimental adjustments for the phase-mostly and amplitude modulation LCTV's is presented. Optical Fresnel and Fourier complex amplitude holograms and optical complex amplitude correlation filters are shown.

The reflective magneto-optic spatial light modulator (R-MOSLM) device was developed over the past few years by a group of Litton divisions and Carnegie Mellon University. The device has been fabricated into 128 X 128 arrays on 24 micron pitch. The performance of individual devices has been reported in previous years. This paper describes the use of the device in an optical correlator. Litton has been developing the miniature ruggedized optical correlator (MROC) for use in a variety of pattern recognition applications. This paper discusses the packaging of the device, the drive electronics, and the interfacing of the device to the MROC unit.

Theory concerning the use of liquid crystal televisions as phase modulators in 4f correlators is presented. A computer simulation of a 4f correlator is presented. The simulation depicts the response of a 4f system to matched and phase-only filters calculated for amplitude- and phase-only inputs,

A new drive circuit for the liquid crystal display (LCD) of the InFocus TVT-6000 video projector is currently under development at the U.S. Army Missile Command. The new circuit will allow individual pixel control of the LCD and increase the frame rate by a factor of two while yielding a major reduction in space and power requirements. This paper will discuss results of the effort to date.

The optimum receiver for pattern recognition is designed based on multiple-alternative hypothesis testing with unknown parameters for detecting and locating a noisy or a noise-free target in scene noise that is spatially nonoverlapping with this target. The optimum receiver designed for a noise-free target has the interesting property of detecting with no error a noise-free target having unknown illumination using operations that are independent of the scene noise statistics. In this paper, we investigate the performance of the optimum receiver designed for nonoverlapping target and scene noise in terms of rotation- and scale-sensitivity of the input targets, and discrimination against similar objects. Since, in practical systems, it is not possible to have a completely noise-free target, we examine how the performance of the optimum receiver designed for a noise-free target is affected when there is some overlapping noise on the target.

A modulator using a GaAs/AlGaAs MQW structure has been designed and grown to modulate phase in the 850 nm region.