This PDF file contains the editorial “Culture Clash” for OE Vol. 37 Issue 07

The U.S. Army Tank-Automotive Research, Development and Engineering Center (TARDEC) has had a broad interest in modeling and simulation techniques during the last several decades. Specifically, as army ground vehicle designers, our group is interested in predicting the performance of military observers for detecting and discriminating vehicle targets in complex background scenes.

The role of the atmosphere in target acquisition modeling is investigated experimentally. Three models are compared to experimental results measured on the Golan Heights, israel. Concepts considered are atmospheric attenuation versus atmospheric blur, and contrast- limited (blur-limited) versus noise-limited imaging. Results indicate that the role of the atmosphere in target acquisition is blur rather than attenuation and that for ranges of the order of a few kilometers, modern sensors are limited by atmospheric blur rather than by noise. A significant portion of the atmospheric blur derives from small angle forward scattering by aerosols, which actually increases measured temperature differences for ranges up to a few kilometers.

Any image acquired by optical, electro-optical, or electronic means is likely to be degraded by the environment. The resolution of the acquired image depends on the total modulation transfer function (MTF) of the system and the additive noise. Image restoration techniques can improve image resolution significantly; however, as the noise increases, improvements via image processing become more limited because image restoration increases the noise level in the image. We characterize the influence of the MTF and noise level on human target acquisition probability to ascertain the advantages, if any, of image restoration. Conditions when restoration would be advisable are determined.

A target attractiveness model describes field-of-view search as a two-step process. The observer first decides to move to some location based on its attraction and size. These attractive points are further divided by a second decision, which determines whether an observer will perform a detailed examination of the area or will choose a new attractive point. Simple metrics based on preattentive vision processes such as a probability of edge (POE) metric or the peak signal are used to represent the attractiveness. The model is tested on human performance experiments performed by the Night Vision and Electronic Sensors Directorate, which provided eyetracker data for a series of static field-of- view images. The model predicts the statistical properties of the eye- tracker points associated with the saccades, attractive points, and the examinations. The model predicts a clear statistical distinction between these points based on the distribution of the attractive metric in the image. In the simplest case considered, the saccade points are shown to be random with respect to the image (distribution of attractiveness equivalent to whole image) while the attractive points and examinations have distributions weighted by the first and second power of the attractiveness, respectively.

An assessment of human visual performance for detecting low contrast targets is essential in many army target acquisition applications. We examine the comprehensive Blackwell-McCready (BM) contrast threshold data set that was acquired under carefully controlled laboratory conditions. We also examine the applicability of this methodology for all levels of apparent contrast. A hyperbolic numerical curve fit is made on the measured minimal target/background contrast threshold data as a function of target area and ambient background luminance for a wide range of presentation times. An extensive parametric analysis shows that the largest variations in human performance occur during the time intervals before dawn and after dusk. The ABC law generalizes Ricco's and Blackwell's laws and gives some particularly simple relationships between the relative target/background contrast, target area and ambient luminance, which are useful for military and commercial applications.

A prediction of human visual target detection performance is important for many military and civilian applications. A previous paper by the authors in this journal develops an extensive numerical hyperbolic curve fitting routine for the Blackwell-McCready (BM) laboratory contrast threshold data set. The BM data set is extended to real world scenes by developing a new operational definition of ambient luminance. The process involves an empirically derived formula for computing the diurnal ambient luminance as a function of solar position, time of day and observer location on the earth. An analytical methodology predicts probability of detection Pd for various target areas, contrast levels and background ambient luminance. A subsequent analysis shows that Pd is a very sensitive function of target parameters during the twilight region before dawn and after dusk. Several numerical examples illustrate the utility of this formalism to various target detection applications involving aided and unaided optics.

The probability of detection (Pd) of moving targets in visually cluttered scenes is computed using the fuzzy logic approach (FLA). The FLA is presented as a robust method for the computation and prediction of the Pd of targets in cluttered scenes with sparse data. A limited data set of visual imagery is used to model the relationships between several input parameters: the contrast, vehicle camouflage, range, aspect, width, and experimental Pd. The fuzzy and neurofuzzy models provide predicted Pd values that have 0.9 correlation with the experimental Pd's. The results obtained indicate the robustness of the fuzzy-based modeling techniques and the potential applicability of the FLA to those types of problems that have to do with the modeling of aided or unaided detection of a signal (acoustic, electromagnetic) in any spectral regime.

Blackwell described observer performance in terms of three variables: target size, contrast, and overall brightness. However, Blackwell's experiments involved uniform targets and backgrounds. Formulations for effective contrast to enable applying his data to generalized problems have resulted in substantial differences between observed and expected results. Response equivalent delta-T (REDT) is designed to solve this problem. REDT is a contrast metric that is defined in terms of the distance between a signature point and an experimentally derived threshold response surface. It provides more realistic predictions of target detectability than more customary contrast metrics in many situations involving nonuniform targets and backgrounds. In an appropriately controlled experiment, a correlation coefficient of 0.982 is obtained between observer data and REDT values. Although REDT has not been demonstrated to significantly outperform other contrast metrics in realistic applications, the scatter in such data can be understood in terms of uncontrolled variations in clutter, target size, and second-order image statistics, which describe texture. Gray-level cooccurrence matrices provide a plausible tool for understanding situations in which REDT fails to perform well. REDT is most appropriate when targets are small, resolution is low, and clutter is moderate to high.

A simple and efficient psychophysical procedure is presented to quantify the visual conspicuity of a target in a complex (natural) scene. Measurements can easily and quickly be performed in the field or in complex environments. Only a few observers (typically two to three) are needed to achieve sufficient accuracy. We show that this conspicuity measure predicts human visual search performance in realistic and military relevant complex scenarios. Also, conspicuity measured on photographic slides agrees with conspicuity measured in the field, which implies that the new conspicuity measure can be used in combination with photosimulation studies to optimize and evaluate the visual distinctness of displayed information.

The characterization of electro-optical system performance by means of the minimum resolvable temperature difference (MRTD) or the minimum resolvable contrast (MRC) has at least three serious disadvantages: (1) the bar pattern stimulus is theoretically and practically unsuitable for 1-D or 2-D spatially sampled systems such as pixel-array cameras, (2) spatial phase is not taken into account, and (3) the results depend on the observer’s subjective decision criterion. We propose an adequate and easily applicable alternative: the triangle orientation discrimination (TOD) threshold. The TOD is based on an improved test pattern, a better defined observer task, and a solid psychophysical measurement procedure. The method has a large number of theoretical and practical advantages: it is suitable for pixel-array cameras, scanning systems and other electro-optical and optical imaging systems in both the thermal and visual domains, it has a close relationship to real target acquisition, and the observer task is easy. The results are free from observer bias and allow statistical significance tests. The method lends itself very well to automatic measurement, and can be extended for future sensor systems that include advanced image processing. The TOD curve can be implemented easily in a target acquisition (TA) model such as ACQUIRE. An observer performance study with real targets shows that the TOD curve better predicts TA performance than the MRC does. The method has been implemented succesfully in a thermal imager field test apparatus called the thermal imager performance indicator (TIPI) and may be implemented in current MRTD test equipment with little effort.

In an extensive observer experiment, identification and classification performance with a CCD camera system was measured for ship targets. The experiment was specifically designed for target acquisition (TA) model validation and improvement. Target type, target aspect angle, viewing depression angle, target contrast and range were varied systematically, and 50% correct acquisition ranges were determined. The observer data were used to validate several basic assumptions in the TA model ACQUIRE and to test whether the recently developed triangle orientation discrimination (TOD) system performance characterization curve (Bijl and Valeton, 1998), when implemented in ACQUIRE, leads to performance predictions similar to or better than the minimum resolvable contrast (MRC) does. The results of the study are (1) the TOD curve predicts the effect of target contrast very well while the MRC curve is too steep, thus proving the value of characterizing a sensor with the new procedure; (2) ACQUIRE predicts the effect of target orientation on acquisition range quite well, using the square-root area of the target as characteristic size; (3) the Johnson cycle criteria for identification and classification of sea targets are 9 and 4, respectively, which means that angular dimensions for sea targets must be 1.5 to 2 times larger than for ground targets to obtain the same level of acquisition performance; (4) similar TOD acquisition criteria were defined, and these appeared to be quantitatively the same as the Johnson criteria; and (5) in 95% of the conditions, the ratio between predicted and measured range falls between 0.5 and 2.0 when the MRC is used, and between 0.6 and 1.6 when the TOD is used. Overall, considering the simplicity of ACQUIRE, its predictions are reasonable, especially when the TOD is used.

A new computational visual distinctness metric based on principles of the early human visual system is presented. The metric is applied to quantify (1) the visual distinctness of targets in complex natural scenes and (2) the perceptual differences between compressed and uncompressed images. The new metric is shown (1) to predict human observer performance in search and detection tasks on complex natural imagery, (2) to correlate with visual target distinctness estimated by human observers, and (3) to correlate with the quality factor of Joint Photographic Experts Group (JPEG) compressed imagery.

Most current artificial vision systems lack robustness and are applicable only to a narrow range of tasks. Crevier (1997) has suggested that this is due to their reliance on a small number of vision mechanisms and to lack of knowledge about how vision algorithms should be integrated. We suggest a systems approach to artificial vision based on computational vision research. The capabilities of biological vision systems are contrasted with those of current piecemeal approaches to artificial vision. A mature, comprehensive vision system, called the Georgia Tech Vision (GTV) simulation is described. GTV incorporates quasilinear filter mechanisms to simulate the processing performed by simple and complex cortical cells. The outputs of these mechanisms are adaptively combined to discriminate targets from clutter and/or one another. GTV outputs predictions of human search and detection performance and/or targeting metrics for automatic target recognition (ATR) applications. Studies validating GTV as a model of human search and detection performance and demonstrating its performance as an ATR are presented.

A comparison between metrics based on computational vision (CV) and principal component analysis (PCA) is been performed. A CV metric is developed based on the response of the CAMAELEON model and compared with a PCA metric on the basis of synthetic aperture radar (SAR) target chip classification. The two techniques are not correlated and are, to some degree, independent. The independence of these techniques could be used to enhance the decision process in aided target recognition (ATR) applications. In addition, it appears that the use of PCA gives a simple way to detect the presence of a target. The evidence indicates that the method used here could be used as a sophisticated window filter to find regions of interest in images.

The problem of target acquisition is considered to be a very involved process and is a serious challenge for the researchers. For several applications of target acquisition, it is worthwhile to compare the compressed and uncompressed images and the perceptual difference between the two images is also significant. A new neural network technique of image modeling by 2-D random fields formulated in the form of an autoregressive moving average process driven by input white Gaussian noise with known statistics is presented. The proposed technique consists of two stages: (1) estimating the parameters of the model and (2) regeneration of the image with the knowledge of the model, its parameters, initial conditions, and white noise. The problem of estimating the model parameters is formulated as an optimization problem solved by a single-layer neural network. Once the model parameters have been estimated as the adaptive weights of the network, the second stage reconstructs the picture from the model. This stage consists of recursively constructing the image using the initial conditions of the original image, the parameters of the model, and white Gaussian noise. Due to the adaptive nature and the computational capability of the neural network, a high-quality image is obtained with this approach. The proposed algorithm reduces the computational complexity and is recommended for the on-line image compression required in target-acquisition-type applications. As the image is constructed using fewer pixel values of the given image in the form of initial conditions, and a few parameters of the model, very effective image compression is achieved. Several computer simulation examples are included to illustrate the effectiveness of the proposed technique.

We propose a broad-based target classifier that recognizes a target on the principle of pattern matching based on associative memory, and can be implemented in hardware with standard complementary metal-oxide semiconductor (CMOS) cells. A major drawback in the field of neural networks is the inability to implement network designs in hardware inexpensively. Here we have designed an associative RAM-net memory neural classifier, which is based on the associative memory model of a winner-take-all classifier. In this model extensive use of ordinary random access memory is made. The benefit of this approach is that the entire architecture can be designed with current CMOS standard cells application-specific integrated circuit (ASIC) technologies as opposed to current analog very large scale integration (VLSI) approaches. Since RAM can be easily added using the standard cell ASIC approach, a low-cost implementation for a wide variety of neural classification problems is provided. The integration of all necessary features on a single chip results in an easily implementable, low-chip-count classification system. A full gate-level design of the architecture is created and simulated using the VHDL hardware programming language (Navabi, 1993; Mentor Graphics, 1992). The results are very encouraging, and warrant further research into various memory types and configurations.

The statistical characterization of realistic, complex backgrounds where targets may be embedded is essential in the optimization of methods for target acquisition. A modeling framework for complex backgrounds that yields perceptually realistic images may provide a path towards such essential characterizations. To this end, we developed a framework for the synthesis of statistical textured backgrounds. The results of the syntheses of three rock-like structures and that of grass indicate some of the capabilities of the framework. We further extended the methods to synthesize biological tissue samples which present two forms of background complexity: a slowly, spatially varying meanbackground and a residual texture image. The extended framework allows synthesis of each component independently. A mathematical phantom for modeling inhomogeneous backgrounds is then proposed. First and second order statistics of various textures are then presented and a measure of distance between two images is proposed. Finally we discuss how such a framework may lead to effective statistical descriptions of such complex backgrounds for target acquisition.

The Hough transform is suggested in literature as an effective technique for track initiation. However, to date, most papers have focused on simplistic applications of this technique based on one sensor, one target, and no clutter scenarios. We present a detailed methodology applying the Hough transform to the multiple target, multiple sensor data association and track initiation problem. Using a two-tiered methodology, bearing and range tracks are detected in separate Hough spaces. The measurements defining these tracks are subsequently associated for the purpose of track detection and initiation. The effectiveness of this technique is demonstrated through several comprehensive simulation scenarios.

An important step in target acquisition is to be able to label various unconnected objects (components) in the scene. We present new algorithms for labeling connected components in a binary image. Eight connectivity for object and background pixels is assumed. The sequential algorithm is described first. The computational complexity of the algorithm is linear (optimal) in the number of object pixels in the image. The memory requirement is also linear in the number of object pixels. The representation of the connected components used in the algorithm makes it easy to calculate certain properties of regions, i.e., area, perimeter, etc. The algorithm is then parallelized and implemented on a shared memory computer. The computational complexity of this parallel algorithm is O(é\log n?) for an image with n object pixels using n processors, and is the best achieved yet. The implementations of the algorithm on several distributed memory architectures, i.e., a binary tree connected computer [O(log n)], a unidirectional and bidirectional mesh connected computer [O(n1/2)], and a hypercube computer [O(log n)] are discussed. The computational and communicational complexities of these implementations are computed, and are the best yet achieved. The algorithm is easily extended for gray-level images without affecting the complexities. Results of the implementation on the sequent balance multiprocessor are presented.

The sensitivity of photodetectors (vidicon type tubes with Se+Sb2S3, Sb2S3, and PbO + PbS phototargets), used in applied TV systems, is estimated. The sensitivities of these photodetectors rated in photometric units (lux) show differences of three orders of magnitude, whereas in radiant units (Wcm-2), the sensitivities of these photodetectors differ only several times. This effect results from the fact that the spectral characteristics of the sensitivities of the investigated photodetectors in the region from 0.38 to 0.67 µm (the region of the measurement of illuminance that changes in photometric units) differ significantly. The amount of radiation (in radiant units) required for photodetectors with different types of phototargets to generate the same output signal appears practically the same.

The temperature and electromagnetic effects on erbiumdoped fiber (EDF) amplifier (EDFA) systems are investigated. First, the temperature sensitivity at the connection of single-mode fiber and erbium-doped fiber in an EDFA system pumped at 980 nm amounts to -0.023 dB °C21 (or -7.2 mV °C-1) in the temperature range of 25 to 95°C. This can be accounted for by the temperature-induced coupling loss and the gain characteristics of the EDFA. Second, the effect of electric and magnetic fields on the EDF of the EDFA system are shown. The electromagnetic interference effect on the photodiode and its light power and wavelength dependences are also discussed.

By inserting a bidirectional erbium-doped fiber amplifier (EDFA) in between an optical circulator and a fiber Bragg grating (FBG), we realize an FBG-integrated optical limiting amplifier (OLA) with high dynamic range. The dual-pass OLA has a wide dynamic range of over 40 dB and a saturation signal output power of about 13.0 dBm. The performance of dual-pass OLA has no obvious degradation due to back reflection of the amplified signal. A negligible power penalty of about 0.3 dB is observed when compared with other conventional configurations. The FBG-integrated OLA configuration has potential application in wavelength division multiplexing (WDM) systems where high saturated power is needed for multichannel transmission.

Code division multiple access (CDMA) as a means of multiplexing in optical fiber sensor networks has been experimentally studied. Crosstalk levels between channels below -60 dB were demonstrated in a reflective network for Fabry-Perot interferometers. The limitations of the method with regard to crosstalk are discussed.

Wood selection for musical instrument making is usually based on visual estimation and on manual mechanical and vibrational subjective testing. The method proposed uses TV holography and holographic interferometry with a double-pulsed laser to visualize the standing acoustic waves and the impact generated by transient waves as inspection testing tools. The theory of bending waves in orthotropic plates is used to interpret the experimental results. This enables us to estimate Young’s moduli parallel to and perpendicular to the grain and the in-plane shear modulus in wood. The testing is nondestructive, so the expensive wooden plates can be used in production. Properly modified, the test can be applied to the judgment of any wooden and wooden composite plates. The 3-D charts for a set of the physicoacoustical characteristics—density, modulus of elasticity, and acoustic constant on the axes—are confirmed as a useful procedure that can help to provide more objective results in wood grading for materials selection in the design of musical instruments, mainly bowed string instruments and pianos.

A new frequency modulation/time division multiplexing (FM/ TDM) hybrid reticle for an imaging system is proposed. The spinning reticle modulates the pixels contained in one column of scanned image to different frequencies according to their location. An imaging system using the proposed reticle can reconstruct a high-resolution 2-D image with the same number of bandpass filters as a FM reticle system. An optical experiment shows that the proposed FM/TDM hybrid reticle can be used as a low-cost imaging system.

We develop a method to estimate the 3-D location of straight edges captured from a mobile camera. The location uncertainty is also computed. A new direct formulation is presented that avoids both full optical flow and correspondence computations. The algorithm uses image regions that support straight edges and can be applied to small, known camera motions. When the image disparity is small, coherent depth results are obtained in a way more efficient than with a correspondence-based approach.

Scaled versions of a variety of silicon light-emitting diode elements (Si LEDs) have been realized using a standard 1.2-µm, doublepolysilicon, double-metal, n-well CMOS fabrication process. The devices operated with a n+p junction biased in the avalanche breakdown mode and were realized by using standard features of the ORBIT FORESIGHT design rules. The elements emit optical radiation in a broad band in the 450- to 850-nm range. An emitted intensity (radiant exitance) of up to 7.1 µW/cm2 (or about 8 nW per 60-µm-diam chip area) has been obtained with 5 mA of current at an operating voltage of 18.5 V. Excellent uniformity in emission intensity of better than 1% variation was obtained over areas as large as 100x500 µm. A best power conversion efficiency of 8.7x 10-8 and a quantum efficiency of 7.8x 10-7 were measured. All of these values are about one order of magnitude better than previously reported values for Si LED avalanche devices. Coupling between the elements as well as electro-optical coupling between an element and an optical fiber was realized.

Apertureless Scanning Near-field Optical Microscope (SNOM) receives an increasing interest in local imaging and analysis. We report a hybrid microscope composed of a commercial Atomic Force Microscope (AFM) and an apertureless SNOM, which operates both in reflection and transmission modes with several illumination and collection systems. The optical probe is a commercial AFM tip integrated on a silicon cantilever. The AFM is operated in the intermittent contact mode at the resonance frequency of the cantilever. We present the images obtained on a grating of cylindrical dots of aluminum (diameter is 200 nm, height is 20 nm) and we show the effects of some optical parameters (polarization, direction of illumination and collection) on the SNOM images.

The outcomes of some experimental and simulation studies performed on an incoherent-erasure photorefractive joint transform correlator (PR JTC) in the two-wave- and four-wave-mixing geometries are demonstrated. Computer simulation results show the effectiveness with which nonlinear transfer characteristics can be achieved in the incoherent-erasure configuration of the JTC. These can be exploited to attain different filter operations, e.g., matched, phase-only, and inverse filters by varying beam ratios. Experimental limitations with PR barium titanate crystal entail the incoherent erasure (IE) JTC operation with all filter actions and encourage to this goal further investigations with fast PR materials.

A simple setup that is capable of performing the function of an optical XOR logic gate is proposed. The setup consists of two BaTiO3 self-pumped phase conjugators as its principal elements. The mechanism of the gate is based on the phenomenon of self-pumped phase conjugation and the suppression of it by ordinarily polarized beams. In this way, the two input channels of the system are not required to be mutually coherent. In addition to its use as an XOR gate, the setup can be used as an image subtractor. And, as in the case in which it is used as a logic gate, the two input images do not need to be mutually coherent.

Interferometry has always been a powerful tool to diagnose the response of liquids when changes of status parameters (e.g., temperature or concentration) induce modifications in their optical properties. interferometric measurements are based on the ability to measure variations in the optical path length around a reference configuration. investigations done so far on heat convection driven by capillary forces indicate that the observation of both the bulk phase and of the free surface is instrumental for understanding the physical mechanisms steering the heat transfer phenomena in "weightless liquids." When used in space applications, conventional interferometers suffer some fundamental drawbacks because of the severe requirements in terms of the mechanical stability of the optical elements. Holographic interferometry removes the most stringent limitations of classical interferometry, but requires precise positioning of the recording plate, with accuracy better than half a wavelength. The superior feature of an electronic speckle pattern interferometer (ESPI) is that it enables real time correlation fringes to be recorded by a video camera and displayed on a television monitor, without recourse to any form of photographic processing or plate relocation. This comparative ease of operation enables the technique of electronic speckle pattern interferometry to be extended to considerably more complex problems of deformation analysis and measurement of refractive index modulation. Since it basically works as a time differential interferometer, measurements can always be referred to a well known configuration and condition of the test sample, reducing or even eliminating the requirements on mechanical stability. We describe how a double-path ESPI is accommodated within the optical diagnostics of a microgravity payload, fluid physics facility (FluidPac), due for launch in 1999 on the Russian retrievable capsule Foton. The ESPI here described enables one to observe and quantify the deformation of the free surface of a liquid subjected to a thermal gradient. Motions induced by the convective flows in the bulk phase can be monitored at the same time. The main features of the ESPI are presented together with design outlines and optical performances.

The complexity of optical thin film stacks leads to new techniques being developed for characterization of components after deposition. We show the complementarity between optical characterization techniques and grazing x-ray reflectivity. We illustrate the efficiency of that complementarity for the reverse engineering of a 7-layer mirror and a 50-layer bandpass filter.

In a previous paper it was shown that when the Fourier transform technique is applied to binary images (low light level images where the photocurrent distribution is clipped to be binary or one-bit digital images), the processing time necessary for the spectrum estimate is considerably reduced.

This PDF contains the communication section which includes "Gaussian beam intensity flattener"