This PDF file contains the editorial “Service” for OE Vol. 37 Issue 01

The breadth and depth of the advances in recognition techniques are rather broad. The active investigators thus come not only from the optics community but from many other disciplines.

The probability of detection (Pd) of targets in static infrared and 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. The Mamdani/Assilian and Sugeno neuro-fuzzy-based models are investigated. A large set of infrared (IR) imagery and a limited set of visual imagery are used to model the relationships between several input parameters: the contrast, camouflage condition, range, aspect, width, and experimental Pd. The fuzzy and neuro-fuzzy models gave predicted Pd values that had 0.98 correlation to the experimental Pd’s. The results obtained indicate the robustness of the fuzzy-based modeling techniques and the applicability of the FLA to those types of problems having to do with the modeling of human-in-the-loop target detection in any spectral regime.

An optical security system using phase encryption of biometric identification data is demonstrated. For this effort, fingerprints are used as the biometric signature. Two types of phase masks are fabricated: photopolymer film and polyimide film. The masks containing the phase-encrypted fingerprints are tested using a joint transform optical correlator. The results of the tests demonstrate that the encryption methods are successful, as the target fingerprint is recognized and the test print is rejected.

A novel, all-optical joint transform correlator using bacteriorhodopsin (bR) spatial light modulators is presented. This architecture employs relatively inexpensive white light sources and He-Ne lasers to exploit the photoinduced anisotropic properties of the bacteriorhodopsin films. Due to the high spatial resolution of bacteriorhodopsin films, the system offers the possibility of being reduced to a compact size. Advantages and limitations of the system are described and comparisons to previously reported optical correlators using bR films are discussed. The authors believe that this is the first time an all bacteriorhodopsin based optical correlator is reported.

A new implementation of the joint Fourier transform correlation (JFTC) utilizing the phase information in the Fourier plane is proposed. When a phase-based thresholding scheme is used on the joint Fourier spectrum, the computation is significantly accelerated. We obtain almost double normalized correlation peaks with comparable discrimination peak ratio as that of binarized JFTC.

A multireference joint transform correlation technique for rotation-invariant recognition as well as estimation of the rotational distortion of input scene target(s) is proposed. Reference patterns corresponding to different ranges of distortions are synthesized from a set of training patterns. Different types of training set selection strategies are discussed.

The advantage of discrimination sensitivity of a preprocessed binarized joint-transform correlator (PBJTC) has been combined with the rotational tolerance of circular harmonic component (CHC) analysis to realize a JTC which gives a discrimination-sensitive, rotation-invariant pattern recognition system. Both computer simulation and experimental results are provided. The use of gradient preprocessing and adaptive binarization helps in suppressing the sidelobes otherwise associated with a conventional CHC-based JTC. The simulation study shows that such a composite JTC gives a sharp correlation peak for the rotated target and discriminates nontarget, producing a very low crosscorrelation peak that stays below the detection threshold. Although it involves two additional improvisation steps (gradient preprocessing and computation of the CHCs), they can be performed prior to the correlation operation, and the simple binarization process can be realized faster optically along with a dedicated digital comparator.

A technique utilizing an electron-trapping (ET) film to block the extended dc term in a Fourier optical system film is proposed and demonstrated. Since the blocking process is actually based on the subtracting property of the ET film, it is a soft blocking, which differs from, and is better than, simple hard blocking with a stop.

The detection efficiencies for the nonzero-order joint transform correlator (N0JTC) and the conventional joint transform correlator (CJTC) are evaluated. We show that the N0JTC outperforms the CJTC under both the noiseless and noisy conditions, due to the high diffraction efficiency (DE) of the N0JTC. We also show that false alarms created by intertarget modulation can indeed be avoided using the N0JTC.

A simple method of removing the zero order in a joint transform power spectrum, i.e., using the nonzero-order joint transform power spectrum (N0JTPS), is discussed. We show that the nonzero-order joint transform correlator (N0JTC) performs better than the conventional joint transform correlator (CJTC). In short, the N0JTC offers high detection efficiency, better input pixel utilization, and avoidance of false alarms due to multitarget intermodulation. Simulated and experimental demonstrations are provided.

Compansive diffraction nonlinearities reduce the noise and improve the performance of joint transform correlators. The compression and expansion of the photorefractive two-beam coupling parallel optical device is similar to that of the limiting square-law serial electronic receiver. Computer simulations of this device indicate superior performance over a wide range of noise levels.

A new all-optical nonlinear fringe-adjusted joint transform correlation (JTC) technique is proposed. The proposed scheme is based on incoherent-to-coherent conversion with the erasure of a four-wave mixing arrangement in a real-time photorefractive media. This technique is very fast because it generates the correlation output in one step and all intermediate processing is performed in the optical domain. Furthermore, this technique eliminates the problems due to nonlinearities associated with the square-law devices and spatial light modulators used for recording the joint power spectrum. Computer simulation results show that the proposed joint transform correlator yields superior performance when compared to the alternative joint transform correlator architectures.

We discuss an optimum method for target detection where the input data consist of a time series of images in one or more data channels. Backgrounds are assumed to be complicated by nonstationary Gaussian clutter that is correlated over time and across channels and may be further corrupted by highly localized non-Gaussian interference terms that appear target-like. Because of the nonstationary clutter, methods based on the Fourier transform are impractical. Instead, we use a pixel-based autoregressive (AR) model. To deal with the inhomogeneous clutter, we segment the data into locally stationary regions; each region is then whitened using its estimated AR parameters, and an optimum matched filter is applied to the whitened data. The main contributions are the following. First, we develop a generalized AR model to describe multiple frames of multiple-channel data. Second, we introduce a novel automated method for the detection of small targets in nonstationary background. Third, we discuss some special applications such as the detection of small targets in non-Gaussian background clutter. We describe in detail the implementation of these techniques, and demonstrate their performance using both synthetic data and real data obtained from the Compact Airborne Spectrographic Imager (CASI).

The output of an optical correlator is sensitive to rotations of the input object. However, the correlator cannot distinguish the direction of this rotation (clockwise or counterclockwise). We introduce a rotationally sensitive multiplexed filter that senses both the direction and amount of rotation of the input object without employing iterative filter techniques. Our filter simultaneously forms three correlation outputs at three different spatial locations. Each output represents the correlation of the input object with a different rotated version of the reference object. Comparison of the three correlation outputs enables the direction and amount of rotation to be determined. This approach is insensitive to changes in the input object intensity because the various correlation peaks are formed simultaneously. Experimental results show an angular sensitivity of less than 0.5 deg

A computer-generated binary amplitude-compensated matched filter is proposed. The filter has only a few discrete gray levels. As compared with the filters made by the optical method, the performance of filters made by computer-generated holography appears no worse, and fewer discrete gray levels are obtained; it is compatible with an e-beam lithography or an amplitude-modulated spatial light modulator. The excellent performance in discrimination and in Horner efficiency of the amplitude-compensated matched filter is preserved. The circular harmonic version of the amplitude-compensated matched filter is also investigated.

It is important for an optical pattern recognition filter to produce a localized correlation response without significant sidelobe structure. This paper notes, and investigates further, the high degree of robustness of the correlation response of a wideband filter in the face of severe frequency domain modulus disruption. The phenomenon is demonstrated by simulation, and analysis is presented to explain the observations. This property may be of value in filter design by allowing the relaxation of frequency domain constraints in the knowledge that correlation peak structure will not drastically deteriorate.

Several algorithms are developed to produce optical correlation filters for obscured inputs. Each algorithm is optimal for a different set of performance criteria. Computer simulations using binary inputs indicate that some algorithms operate over a wide range of obscurations (down to 5% of the input) while others have restricted operating ranges.

The experimental and theoretical operation of the dc-blocked phase-only filter correlator is studied for obscured inputs. The performance of the nearly optimal filter is compared with the performances of other varieties of simple optimal filters that are capable of real-time implementation with spatial light modulators. Uses of these correlation filters in associative memories is investigated.

In designing optimal correlation filters for implementation on arbitrary spatial light modulators, we need to determine two parameters (a gain and an angle) to maximize the performance criterion of interest. We present a method to find these optimal gain and angle parameters directly.

A distortion-invariant joint transform correlation (JTC), based on the fractional power fringe-adjusted joint transform correlation and synthetic discriminant function, is investigated. By suitably adjusting the parameters used to characterize the fractional power fringe-adjusted filter formulation, we obtain JTC-based classical matched filtering, fringeadjusted filtering, and amplitude-modulated phase-only filtering. Computer simulation results show that the proposed joint transform correlation is distortion-invariant for the target image from the training set.

Preprocessing of the input scene can be used to improve the performance of an optical correlator. The result of the preprocessing algorithms can be bipolar real-valued images, with positive and negative values, for input to the optical correlator. We present different procedures for the implementation of the bipolar real-valued signal with spatial light modulators that work in either amplitude-only or phase-only regimes. We apply these techniques to the implementation of bipolar realvalued preprocessed input images in color pattern recognition.

Many commercial and law enforcement applications of face recognition need to be high-speed and real-time, such as passing through customs quickly while ensuring security. However, face recognition by using computers only is time-consuming due to the intensive calculation task. Recently optical implementations of real-time face recognition have attracted much attention. In this paper, a real-time optoelectronic morphological processor for face recognition is presented. It is based on original-complementary composite encoding hit-or-miss transformation, which combines the foreground and background of an image into a whole. One liquid-crystal display (LCD) panel is used as two realtime SLMs for both stored images and face images to be recognized, which are of 256x3256 pixels. A speed of 40 frames/s and four-channel recognition ability have been achieved. The experimental results have shown that the processor has an accuracy over 90% and error tolerance to rotation up to 8 deg, to noise disturbance up to 25%, and to image loss up to 40%.

Recognition of junctions in scenes containing polyhedral objects has been shown to be an effective means of encoding for modelbased matching. However, noise and inconsistent lighting in the images limit the amount of useful information than can be extracted for junction identification. A wavelet-based system for polyhedral junction recognition that combines the wavelet-derived edges and line elements at multiple levels of resolution to produce a labeled image is introduced. The results of some experimental studies are also reported.

Texture measures offer a means of detecting targets in background clutter that has similar spectral characteristics. Our previous studies demonstrated that the ‘‘fractal signature’’ (a feature set based on the fractal surface area function) is very accurate and robust for grayscale texture classification. This paper introduces a new multichannel texture model that characterizes patterns as 2-D functions in a Besov space. The wavelet-based fractal signature generates an n-dimensional surface, which is used for classification. Results of some experimental studies are presented demonstrating the usefulness of this texture measure.

Based on the modified fractional order Fourier transform (MFROFT), we introduce a modified fractional correlation operation (MFCO). We propose two systems to implement the MFCO experimentally. These are analyzed theoretically to conclude that the MFCO of two objects equals the conventional correlation between the two objects modified by multiplying each by a quadratic phase factor that contains the parameter that determines the fractional order. The MFCO is applied to a rectangular slit analytically, numerically, and experimentally. We show that the MFCO results in superior correlation peaks compared to those of the conventional correlation operation.

The concept of multiresolution analysis on the basis of multichannel filtering in the theory of the early stages of human vision offers suitable methods for pattern recognition. A novel micro-optical multiwavelet element, which has the functions of wavelet filtering, beamsplitting, and self-imaging, has been designed and fabricated for a hybrid texture segmentation processor. All sixteen channel filters process in parallel, using a conventional 2-D correlator. Accordingly, sixteen filtered images sensitive to different scales and orientations are obtained simultaneously as features for texture segmentation. Experimental results and primary applications are presented.

We propose a filter structure formed as the superposition of an impulse removal filter, based on a modified K-nearest-neighbor operation and discrete wavelet transform based noise reduction. This filter removes impulsive and Gaussian noise and mixed noises well. We also suggest an efficient architecture for implementing the impulse removal filter used in this structure.

This paper concerns model-based vision for road traffic scene analysis. A traffic vision system is described. The three main modules of the system are movement detection, vehicle localization and discrimination, and vehicle tracking. This paper outlines our work on the localization and discrimination module. Emphasis is on recovering 3-D poses of road vehicles in given image regions. Two classes of algorithms are described, one based on symbolic image features (line segments), and the other simply on image intensity gradients. A priori knowledge about traffic scenes and vehicles is exploited to improve the performance and efficiency of the algorithms. The algorithms are tested extensively with routine outdoor traffic images, and examples are included to demonstrate their principles.

A spatial Fourier transform approach is employed to investigate the acousto-optic interaction in the near-Bragg regime with cw planar sound and a light beam of arbitrary spatial profile. Four coupled differential equations that can exactly describe the near-Bragg acoustooptic interaction are derived from Maxwell’s equations. A spatial Fourier transform approach is applied to the coupled-wave equations to express the interaction in the spatial frequency domain. The spatial Fourier transform approach used here is identical to the technique for solving the paraxial wave equation to derive the transfer function of propagation and hence the Fresnel diffraction formula during free-space propagation of a light beam in the presence of diffraction. Since there are no analytic solutions for these coupled equations, a well-known numerical method, viz., the Runge-Kutta-Fehlberg, is used to solve the equations. Detailed numerical simulation is performed, which involves spatially Fouriertransforming the input light beam profile to calculate the spectra of the diffracted light beam and hence their profiles in the space, using the inverse transform. Experimental results are also presented to check the validity of our approach.

The principle and design of a cost-effective optical fiber communication system are proposed to support hybrid digital/analog signal transmissions, by simultaneously using pulse-width modulation (PWM) for the analog signal and two-level pulse-amplitude modulation for the digital signal, while essentially requiring the same channel bandwidth as a conventional analog PWM system.

A comparative study of the signal acquisition from straight fiber and tapered optical fibers has been analyzed theoretically for the design of an evanescent wave fiber optic immunosensor. It has been shown that tapered optical fibers have higher sensitivity compared to straight fibers. Among the tapered fibers combination tapered fibers were also found to have higher sensitivity than continuous tapered fibers. Theoretical analysis indicates that an increase in signal is a result of three factors, namely, V number matching, an increase in evanescent absorption by the fluorescent medium, and an increase in fluorescence emission coupling into the fiber. Maximum sensitivity is obtained when the radius of the fiber is reduced to the V number matching radius within 1 cm from the start of the taper with rest of the fiber remaining untapered. Also, a higher signal level is obtained when the launching light fills the numerical aperture determined by the tapering ratio of the fiber.

Light-beam profile deformation during scattering with the photorefractive grating formed by the diffusive, photovoltaic, and drift mechanisms is investigated in the case of anisotropic Bragg diffraction for the incident light with arbitrary spatial beam profile under the framework of a three-dimensional configuration. A set of three-dimensional coupled-wave equations that can properly describe the interaction is derived from Maxwell’s equations. In the wave equations, the permittivity profile, which is perturbed by the diffusive, photovoltaic, and drift mechanisms, can be correctly plotted by employing Kukhtarev’s material model. Using a spatial Fourier transform technique identical to the one used in deriving the spatial transfer function from the paraxial wave equation for free-space propagation, a steady-state solution for the output light beams diffracted by this initially stored grating can be calculated in the spatial frequency domain. The nature of the dependence of beam profiles on the material properties, sample thickness, and incident angle, under the combined effects of propagational diffraction, diffusion, photovoltaicity, and drifting, is presented.

The temperature dependence of the Bragg wavelength of fiber optic gratings was measured at temperatures from 4.2 to 350 K. The measured wavelength dependence at room temperature (293 K) corresponds to a refractive index dependence of dn/dT=9.1×10-6 /°C, which agrees with earlier results for fused silica. However, dn/dT decreases with lower temperatures, declining to 3.5×10-6/°C at liquid nitrogen temperature (77 K), and to approximately zero at liquid helium temperature (4.2 K). The Bragg wavelength of a fiber grating embedded in a composite material displayed the same temperature dependence as a non-embedded grating.

Compact color holograms combined with illumination that reconstruct the images at the ideal angle are fabricated. The edgeillumination method that reconstructs a hologram through an edge of the glass cover plate is applied. The image blur of the hologram reconstructed with white light is theoretically analyzed and quantitatively confirmed by experimental results. It is found that the rainbow hologram is superior to other white-light reconstruction holograms in image sharpness. Based on this analysis, a method for recording edge-illuminated rainbow color holograms is described.

Some imaging systems employ detector arrays that are not sufficiently dense to meet the Nyquist criterion during image acquisition. This is particularly true for many staring infrared imagers. Thus, the full resolution afforded by the optics is not being realized in such a system. This paper presents a technique for estimating a high-resolution image, with reduced aliasing, from a sequence of undersampled rotated and translationally shifted frames. Such an image sequence can be obtained if an imager is mounted on a moving platform, such as an aircraft. Several approaches to this type of problem have been proposed in the literature. Here we extend some of this previous work. In particular, we define an observation model that incorporates knowledge of the optical system and detector array. The high-resolution image estimate is formed by minimizing a regularized cost function based on the observation model. We show that with the proper choice of a tuning parameter, our algorithm exhibits robustness in the presence of noise. We consider both gradient descent and conjugate-gradient optimization procedures to minimize the cost function. Detailed experimental results are provided to illustrate the performance of the proposed algorithm using digital video from an infrared imager.

Conventional interferometers are not always appropriate for measurement of rough or uneven nonoptical surfaces. In this paper, oblique incidence interferometry with a spatial phase-shifting technique is proposed for the solution of this problem. This technique can analyze surface profiles using only one image that contains a high carrier frequency, and a mechanical shifting device that requires high repeatability is not necessary. Therefore, more accurate and quick measurement can be realized. To verify this principle, Poisson’s ratio of brittle materials is determined. The surface deformation due to pure bending is measured so that Poisson’s ratio can be calculated by the relationship between two contour patterns of deformation before and after bending.

An optical resonator with a phase-shifting array end mirror is presented to obtain an output beam of spatially suppressed far-field intensity distribution from a CO2 laser with a slab waveguide gain medium. A highly spatially suppressed single-peak intensity distribution in the far field and a power of 46 W have been obtained from a 200-mm-gainlength device.

Rapid generation of large area polished optical surfaces by high-power CO2 laser irradiation is shown. Results focus on glasses with high expansion coefficients (??10-5/°C) conventionally used in the optical industry. The technique involves active beam integration to obtain an intensity irradiation profile with a good uniformity over large spot sizes and is applied to preheated glass samples with initial rms roughness up to 500 nm. To find out the conditions for successful and reliable use of the proposed laser polishing method, the laser-driven heating process was monitored by means of the surface and depth temperature distributions. Whereas the former was determined in situ from the IR radiation emitted by the glass surface, the latter was obtained by comparison of the IR radiation emitted by the bulk sample with a theoretical model. Laser polishing of 5000 mm2 glass surfaces is reported and processes involved in the modification of the surface texture of the irradiated samples are described and discussed.

An important source of error when estimating the 3-D position of a point from two (stereo), three (trinocular), or more cameras is that of quantization error on the image planes. In this paper, we are concerned with bounding the quantization errors when using multiple cameras defined in terms of uncertainty regions in 3-D. We use a geometric error analysis method that models the quantization error as projected pyramids and the uncertainty region as an ellipsoid around the polyhedron intersection of the pyramids. We present a computational technique for determining the uncertainty ellipsoid for an arbitrary number of cameras. A numerical measure of uncertainty bound such as the volume of the ellipsoid can then be computed for aiding camera placement, trajectory planning, and various other multiple camera applications.

One of the basic tasks in developing a multi-wavelength machine vision system for on-line fruit sorting is to determine the sorting model and system parameters including the wavelength (or filter) combination and the sorting threshold. A fuzzy method was proposed to evaluate the performance of a dual-wavelength vision system developed for on-line sorting of apples. Three sorting models were developed based on the sizes of defect blobs extracted from apple images. Twelve dual-filter combinations were tested and compared under different sorting requirements. Experimental results showed the effectiveness of the fuzzy method in determining the model and system parameters for the on-line dual-wavelength vision sorting system to meet any user-specified sorting requirements for both good and defective apples.

We analyze the factors affecting the dynamic range of an automated visual web inspection system and characterize the dependence of the image quality on the performance of the imaging unit. The objective is to form a basis for the systematic design and evaluation of imaging in web inspection. The major noise sources are reviewed and a simple method for the determination of the noise generated in the imaging unit of a web inspection system is proposed. In addition to this, the effects of the nonlinearity in the system response and of the optical characteristics to the attainable dynamic range are discussed. In the paper several measured characteristics of exemplary web inspection imaging units are presented. The results of these measurements sketch the order of the effect of various factors on the dynamic range in actual inspection applications. Finally, an analysis of the dynamic range in a paper inspection application is presented in order to illustrate the relationship between the properties of the imaging unit and the performance of the inspection system.

The demonstration of an optoelectronic sorter unit is presented. This sorter is constructed using a recirculating architecture that passes data optically between two smart-pixel arrays. Each array is implemented using a hybrid combination of CMOS logic and vertical cavity surface emitting lasers (VCSELs). Since implementation of this system is intended as a tool for evaluating the technological issues associated with CMOS/VCSEL smart-pixel device, the system implementation is designed to minimize interconnect complexity while focusing on the smart-pixel design and analysis. A power budget analysis indicates that the photoreceivers require a minimum of ?130?W of optical input power, which leads to a VCSEL power requirement of '415 mW. The frequency response of the photoreceivers and logic circuit reveals that the photodetector limits the operation to ?3 MHz, which is an order of magnitude slower than the logic circuits frequency limit.

Subband decomposition wherein each subimage contains information about a specific range of scales and orientations provides an important vehicle for discriminating between objects with coherent structure and the random patterns of background noise. We use the cortex transform to create a 32-element feature space representing four scales and eight orientations. Simple hypothesis testing based on variances of the feature elements provides an initial indication of the presence and location of structure whose scale and orientation make it stand out from the background. The feature vectors of these somewhat obvious object pixels are then used in a novel training scheme to derive a linear classifier for subsequent recognition of faint, low-contrast objects. The method is used to detect calcification structure in digitized mammograms, and the results are compared with those of a previously published wavelet technique. Measurable improvements in detector output suggest that the new approach has potential.

High-resolution moire´ photography1,2 has been of interest for many years for the investigation of in-plane displacements over a large range of specimen dimensions. The method’s main advantages stem from its requirement to illuminate the object with white light, its user-friendliness in non-laboratory environments, and its ability to provide whole field visualization of the undergone deformations. Optical measurements are, however, available only in the form of displacement and not strain maps, a quantity often of more interest to the engineer.

This PDF contains the communications section which includes "Adaptive termination rule for iterative image restoration"