The U.S. Army and the U.S. Air Force are investigating laser range-gated shortwave infrared (LRG-SWIR) imaging systems for use in target identification. When coupled to an electron-bombarded CCD, the imaging system can obtain high- resolution images at long ranges. Speckle, an image artifact inherent in laser illuminated imaging systems, results from interference patterns caused by the coherent illumination. Laser speckle degrades target identification performance but can be reduced by averaging successive LRG-SWIR images. This research is a first attempt at quantifying target identification performance degradation associated with laser speckle. The research begins with a laboratory experiment to verify a speckle model that includes power spectral density and intensity probability density functions. An LRG-SWIR sensor simulation is developed that includes coherent illumination resulting in speckle target images. A field demonstration is performed to verify the fidelity of the simulation. The simulation is then applied to the NVESD target identification set with various levels of image averaging and blur. Observer performance results are analyzed in terms of target identification probability and the effects of various levels of blur and speckle are characterized.

The windows version of the Night Vision Thermal Imaging System Performance Model, NVTherm, was released in March 2001. NVTherm provides accurate predictions of sensor performance for both well-sampled and undersampled thermal imagers. Since its initial fielding in March 2001, a number of improvements have been implemented. The most significant improvements are: (1) the addition of atmospheric turbulence blurring effects, (2) National Imagery Interpretability Rating System (NIIRS) estimates, (3) and the option for slant-path MODTRAN transmission. This paper presents these modifications, as well as a brief description of some of the minor changes and improvements that have been completed over the past year. These significant changes were released in January 2002.

Display artifacts such as raster and square pixelization associated with flat panel displays can be characterized using sampled imaging systems analysis. When the output raster/pixelization signal is large compared to the image modulation, the overall system performance is degraded. In this research, we investigated display methods for sampled imaging systems, including pixel replication, bilinear interpolation, and a higher-order interpolation. Problems with the performance modeling of these processes are discussed and a perception test is implemented for comparison.

Recent experiments performed at the U.S. Army Night Vision and Electronic Sensors Directorate (NVESD) provide significant insight into the validation of synthetic imagery for use in human perception experiments. This paper documents the procedures and results of target identification (ID) experiments using real and synthetic thermal imagery. Real imagery representing notional first generation and advanced scanning sensor systems was obtained. Parameters derived from the sensor data were used to generate synthetic imagery using the NVESD Paint the Night simulation. Both image sets were then used in a target identification experiment with trained human observers. Perception test results were analyzed and compared with metrics derived from the imagery. Several parameters missing from the original truth data were found to correlate with differences in the perception data. Synthetic data were regenerated using these additional parameters. A subsequent perception experiment confirmed the importance of these parameters, and a good match was obtained between real and synthetic imagery. While the techniques used in this series of experiments do not constitute a definitive method for validating synthetic imagery, they point to some important observations on validation. The main observation is that both target and local background characteristics must be sufficiently specified in the truth data in order to obtain good agreement between synthetic and real data. The paper concludes with suggestions as to the level of detail necessary for truth data when using synthetic imagery in perception experiments.

Human perception tests have been performed by Night Vision Electronic Sensor Directorate (NVESD) addressing the process of searching an image with the intent of detecting a target of military importance. Experiments were performed using both real thermal imagery and synthetic imagery generated using 'Paint the Night' simulation. It was demonstrated that trained observers acquire targets much more quickly than previously expected. This insight was gained by changing the instructions the observers were given for the test. Rather than telling the observers that they were being timed, the observers were given a time limit, as short as 3 seconds. When time limited, the observers found targets quicker. Although false alarms per second increased with the shorter time limits, the ratio of false alarms to detected targets did not increase. A modification to the traditional NVESD search model has been developed and incorporated into Army war games and simulations.

At the moment there are three potential models and associated measures to replace the FLIR92 model and the MRTD as the standard for sensor performance characterization and TA range predictions. These are: (1) the NVTherm model that calculates the MRTD-ID, (2) the TRM3 model that calculates the MTDP and (3) the TOD model that calculates the TOD. The three models are grossly different in theoretical approach. We ran the models for the same set of hypothetical cameras. As default sensor, we used a 'typical' under-sampled Focal Plane Array camera. Then, we independently varied the pre- and post-filter MTF's over a wide range while keeping the sampling frequency fixed so that the cameras ranged between well-sampled and highly under-sampled. The differences in outcome are striking. For example, for a range of sensors around the default the TOD model predicts that performance is determined primarily by the sampling frequency, while NVTherm predicts that pre- an post-filter blur dominate in this region. The results with TRM3 are surprising: the model predicts that increasing image information content by applying microscan to the default sensor, decreases predicted TA performance. The origin lies in the use of the periodic four-bar test pattern and the definition of the MTDP. In the low contrast region, the MTDP and the TOD are similar but the MRTD-ID deviates from these two. The MTDP and MRTD-ID are different even in the well-sampled region where they both should be equivalent to the conventional MRTD. The study shows that the choice of a model has a large impact on sensor design decisions or trade-offs between sensors. The TOD method is the most general of the three approaches and has the strongest basis. Until now, validation of the model predictions is very limited. Therefore, a joint TA performance study with simulated sensors (e.g. those used in the present study) and target contrast as variables would be very useful.

The U.S. Army is currently investigating the differences between various bands in the midwave and long wave infrared spectrum. A holistic approach to quantifying scene information is used in previous research. That is, both natural backgrounds and vehicles are present in scenes when correlation analyses are performed. Similar research has also been performed using hyperspectral imagers. Hyperspectral imagers inherently have poor signal-to-noise ratio (SNR). In this research, a mid-wave infrared broadband sensor was cold-filtered to provide four sub-bands in the mid wave region. A multi-waveband sensor as used to collect midwave infrared imagery of military vehicles and natural backgrounds. Three blackbody sources were placed at the same range as the vehicles for radiometric calibration. The goals were to collect radiometrically corrected data of various targets and process this data for comparative analysis. The images were segmented to remove all unwanted imagery from the images under observation. Correlations were performed to assess the differences in information content.

Infrared Imaging sensors operating in the 3 - 5 um (MW) and 8 - 12 um (LW) spectral bands have long since been traded against one another with respect to mission utility, sensor performance, and system viability (cost factors, technology maturity, etc.). Over the past decade, staring InSb detectors have matured to a refined level (high performance, moderate cost, high yields) and have been used extensively by IR sensor integrators throughout the industry. By the same token, 2-D LW QWIP-based FPA's are fast becoming a viable alternative to traditional LW-scanned technology systems, offering the benefits of mid and large format staring sensor resolution with good sensitivity (even for modest optical F/#'s). With the commercialization of QWIP technology, system viability is rapidly increasing, revealing the need for serious system trade assessments and field measurements to enable the best use of this emerging, complementary detector technology. This paper presents a top-level technical comparison of these two sensor technologies and their use in surveillance/night vision system applications. A variety of technical considerations are discussed to help end users be cognizant of the extent of the trade space that exists between MW and LW staring sensor selection with specific focus on performance comparisons for small, compact militarized IR thermal imaging sensors (including handheld, man-portable and small gimbal products) employing each detector technology in context to various surveillance missions. Application areas include: ground, airborne, and maritime surveillance. Field data is also provided to support the conclusions drawn from these comparisons.

Aliasing characterization is a critical problem in the description of pixelated image acquisition system such as focal plane array cameras for infrared image-forming systems. The aliased spatial frequency spectrum contains only those frequencies below the Nyquist limit. The higher- than- Nyquist frequency components are aliased onto lower- than-Nyquist frequencies. This effect can be described by means of a matricial transformation that is a folded version of the non-aliased transfer function of the system. This matricial analysis helps to understand the effect of the sampling. The transformation can be related with an aliased Modulation Transfer Function (MTF). Several examples of the application of the method are presented along with the description of the matricial formalism.

The Low Cost Microsensors (LCMS) Program recently demonstrated state-of-the-art imagery in a long-range infrared (IR) sensor built upon an uncooled vanadium oxide (VOx) 640 X 480 format focal plane array (FPA) engine. The 640 X 480 sensor is applicable to long-range surveillance and targeting missions. The intent of this DUS&T effort is to further reduce the cost, weight, and power of uncooled IR sensors, and to increase the capability of these sensors, thereby expanding their applicability to military and commercial markets never before addressed by thermal imaging.

The Aviation and Missile Research, Development and Engineering Center (AMRDEC) of the U.S. Army Aviation and Missile Command (AMCOM) conducted a series of Captive Flight Tests (CFT) gathering urban Laser Radar (LADAR) imagery at the McKenna Military Operations in Urban Terrain (MOUT) facility located at Fort Benning, Georgia, July 18 through August 4, 2001.

Although UAV systems have received much interest over the last few years, much of this has focused on either relatively large platforms with complex on-board equipment, or micro systems (typically 6' in every dimension). The operational use of low-cost lightweight UAVs as over-the- hill reconnaissance systems is a new concept offering additional flexibility, providing local knowledge and helping maintain operational tempo. An extensive modeling trade-off study has been performed for different sensor technologies and combinations. The model considered configurations including cooled and uncooled IR sensors, visible-band CCD sensors and image intensifiers. These mathematical models provide an evaluation of sensor performance for both navigation and the gathering of reconnaissance imagery, through Resolution Elements calculations (Johnson criteria) and Signal-to-Noise Ratios. Based upon this analysis, a system specification is presented that exploits next generation sensor technologies. Results obtained from a number of UAV trials are reported and used in order to provide model verification and validation of both the operational concepts and the sensor system modeling activities. Considering the sensor system itself, the low-altitude close-range environment ensures high ground resolved distance and signal-to-noise ratios, with low-cost sensors. Coupled with up-to-date image processing software, the imagery provided directly to the section-level units via a simple standard image interface allows a reduction of time response. Finally, future modeling and trials activities are discussed in the framework of the lightweight UAV system roadmap.

The performance of any scene-adaptive Nonuniformity Correction (NUC) algorithm is fundamentally limited by the quality of the scene-based predicted value of each pixel. TARID-based composite imagery can serve as a scene-based pixel predictor with improved robustness, and reduced noise than that of more common scene-based pixel predictors. These improved properties result in dramatically faster algorithm convergence, generating corrected imagery with reduced spatial noise due to intrinsic nonuniform or inoperative pixels in a Focal Plane Array.

In this paper, we compare two resolution enhancement techniques. The first technique is based on Maximum A Posteriori (MAP) estimation while the second technique performs Temporal Accumulation of Registered Image Data (TARID) followed by Wiener filtering. Both techniques and described and the merits of each one are discussed. Experimental results are presented and conclusions are drawn.

A general method is described to improve the operational resolution of an Electro-Optic (EO) imaging sensor using multiple frames of an image sequence. This method only assumes the constituent video has some ambient motion between the sensor and stationary background, and the optical image is electronically captured and digitally recorded by a staring focal plane detector array. Compared to alternative techniques that may require externally controlled or measured dither motion, this approach offers significantly enhanced operational resolution with substantially relaxed constraints on sensor stabilization, optics, and exposure time.

Imagery data acquired in practice to support tactical surveillance and tracking missions in hostile environments typically suffer from a variety of degradations making it essential to subject the data to digital post-processing aimed at restoration and super-resolution before they can be used for any image exploitation tasks (visualization, target detection and characterization, etc.). A number of novel iterative techniques for resolution enhancement and super- resolution are presently being developed. In this paper, we shall outline two popular avenues that have been followed for constructing iterative processing algorithms, viz. statistical optimization and set-theoretic estimation, and discuss the super-resolution abilities of a powerful algorithm developed from a hybridization of the two approaches. Performance of this algorithm in enhancing resolution in tactical imagery data is illustrated. Notwithstanding the significant resolution enhancement that is possible with these algorithms, the spectrum extrapolation that is central to super-resolution comes only as a by-product and needs to be checked only after completion of the processing steps in order to ensure that an expansion of the image bandwidth has indeed occurred. To overcome this limitation, a new approach of mathematically extrapolating the image spectrum and using it as a constraint in a Projection-Onto-Convex-Sets (POCS) estimation framework to construct an iterative processing algorithm with guaranteed levels of super-resolution is presented.

A new iterative algorithm (EMLS) via the expectation maximization method is derived for extrapolating a non- negative object function from noisy, diffraction blurred image data. The algorithm has the following desirable attributes; fast convergence is attained for high frequency object components, is less sensitive to constraint parameters, and will accommodate randomly missing data. Speed and convergence results are presented. Field test imagery was obtained with a passive millimeter wave imaging sensor having a 30.5 cm aperture. The algorithm was implemented and tested in near real time using field test imagery. Theoretical results and experimental results using the field test imagery will be compared using an effective aperture measure of resolution increase. The effective aperture measure, based on examination of the edge-spread function, will be detailed.

As integrated electro-optical sensor payloads (multi- sensors) comprised of infrared imagers, visible imagers, and lasers advance in performance, the tests and testing methods must also advance in order to fully evaluate them. Future operational requirements will require integrated sensor payloads to perform missions at further ranges and with increased targeting accuracy. In order to meet these requirements sensors will require advanced imaging algorithms, advanced tracking capability, high-powered lasers, and high-resolution imagers. To meet the U.S. Navy's testing requirements of such multi-sensors, the test and evaluation group in the Night Vision and Chemical Biological Warfare Department at NAVSEA Crane is developing automated testing methods, and improved tests to evaluate imaging algorithms, and procuring advanced testing hardware to measure high resolution imagers and line of sight stabilization of targeting systems. This paper addresses: descriptions of the multi-sensor payloads tested, testing methods used and under development, and the different types of testing hardware and specific payload tests that are being developed and used at NAVSEA Crane.

The derivative of the detected incidence in a wavelength interval with respect to temperature has previously been shown to include two terms. The first term depends on the change in the blackbody emission with temperature. The second term depends on the change of emissivity with temperature. The error of neglecting the second terms is analyzed and evaluated for a standard radiation source, a tungsten lamp, for which the measured emissivity data have been published. In this specific case the error changes form a negligible 6% to a significant value of more than 45%.

An hyperspectral imager capable of sensing from 1 to 12 micrometers with three (3) possible field-of-views (FOV) steerable within a field-of-regard eight (8) times larger than the FOV is presented. This level of flexibility imposes several constraints on the front-end optics especially when the maximum etendue of the spectrometer must be maintained for all configurations. This paper presents the design approach and trade-offs leading to a high performance optical design. Other constraints such as mass and volume are also considered. An important limiting factor is the size of the window and its minimum distance to the primary mirror of the telescope. The design has been optimized by re-imaging the aperture stop on each component that are critical in size: the interferometer corner cubes, the steering mirror and the primary mirror of the telescope. A set of two (2) telescopes and two (2) afocal relays are interchanged to produce 3 FOVs with optimized etendue and minimum size on critical components.

In engagement scenarios increasing battlefield emphasis on the trade-off between long stand-off ranges, adverse weather capability and high probability target identification has resulted in the need for an SXGA resolution IR Sensor. Leading on from previous collaborative work with QinetiQ (formerly the UK Defence Evaluation and Research Agency) the UK MoD has awarded a contract, the STAIRS C programme, to Thales Optronics to develop to production such an IR Sensor thus ensuring this leading technology is available to meet the needs of advanced weapon systems and platforms of the future. A UK industry team has been formed to implement an optimisation programme for the productionisation and future applications of STAIRS C modules and the first of a number of UK MoD programmes has selected STAIRS C for a major Air Defence role. The STAIRS C programme has set the demanding requirement of doubling the target identification range of current in-service IR sensors whilst maintaining or improving the situational awareness (Field of View). The programme, technical specification and imaging capability achieved are reviewed in the paper.

The use of uncooled bolometer focal plane arrays in long- wave infrared cameras is becoming more widespread due to reductions in price and increases in sensitivity of these devices. These improvements have made infrared cameras affordable by almost everyone and have expanded their range of usefulness to applications not previously exploited . We describe a compact, lightweight (< 500g) camera that has a 320 X 240 focal plane array with 51 micron pixel size and a 50 milli-Kelvin noise equivalent temperature difference (NETD). Research is being conducted to reduce the pixel size to 25 microns, thus quadrupling the number of pixels, and to reduce the NETD by a factor of two. Recent measurements of the spectral response of this sensor show that it has usable sensitivity out to wavelengths of 25 microns and beyond, so that it would be useful for detecting cold targets and for applications such as detecting weapons concealed beneath clothing, since the longer wavelengths have been shown to penetrate clothing more readily than the generally accepted 8 - 12 micron range of the long-wavelength infrared (LWIR) band. Examples of images obtained with this camera will be shown.

When it was first introduced two years ago, Indigo Systems Corporation's UL3 Alpha, a miniature uncooled infrared camera, set new standards for ultra-low size, weight and power within the thermal imaging industry. Now Omega, the next generation in Indigo's UL3 product line, takes advantage of novel algorithms and packaging concepts to further reduce size, weight, and power while still improving performance. These qualities make Omega an ideal candidate for many commercial and military applications, including fire-fighting, law enforcement, industrial inspection, remote surveillance, miniature unmanned aerial vehicles (UAVs), unmanned ground vehicles (UGV), and numerous other possibilities. This paper describes the design, performance and salient features of the Omega camera. Current and future applications of the UL3 product line are also discussed.

Noise characterization and classification is an important task to evaluate the performance of an infrared imaging system. The focal plane array infrared cameras present several types of noises: fixed pattern noise, 1/f noise, pure temporal noise, etc. The existence of bad pixels showing a singular behavior must be included in the noise description. In this paper we show how the principal component analysis is able to classify the noise of a set of frames into different subsets. The classification method is integrated into a software package that performs the classification of the obtained eigenimages into processes. This method is specially adapted to the analysis of noise in a set of frames because it produces a corresponding set of images characterizing the noise. A result of the analysis provided with this method is the extraction of the fixed pattern noise, the bad pixel identification, the 1/f nosie components and analysis, the pure temporal noise, and some other processes having intermediate time scales.

The development and manufacture of high performance Infrared imaging sensors requires more than just the tools to design and build them it also requires the tools to accurately characterize their electro-optical performance and further utilize this data to better optimize the product as well as monitor many systems in serial production. Santa Barbara Infrared (SBIR) in cooperation with FLIR Systems, Inc. (FLIR) has competed a project to significantly enhance the capabilities of their IRWindows software package, now IRWindows2001, to meet the needs of all levels of IR system developers. This paper will discuss both hardware and software requirements, for IR staring sensor testing and performance evaluation. Key aspects of the new IRWindows2001 software will be described and their utility will be demonstrated with FLIR's MilCAM RECON InSb handheld IR camera.

This paper reviews various passive mm-wave imaging systems. It includes sources of radiation, atmospheric transmission and a brief summary of their various applications. This is followed by a review of methods for detecting mm-wave radiation. From a cost analysis it is shown that scanned systems are at present far more cost effective than focal plane array of receivers for high performance systems. There is a review of the various types of imaging system available with greater emphasis being placed on recent developments and their methods of beam forming. It is concluded that at present, and for the foreseeable future, optical beam forming and beam steering are the most cost effective. Some general remarks are included as to how receivers are matched to their collection apertures, followed by a section on optical beam forming components. The recent development of a lightweight, low cost, high performance reflective lens is included. It is then shown how this may be combined with mechanical scanning systems to form high performance passive mm-wave imaging systems.

Passive millimeter wave imaging has been shown to be a useful for enhanced vision and concealed weapons detection applications. Trex Enterprises is developing a second generation passive millimeter wave imaging system which operates in real time with a 20 X 30 degree field of view and a 2K temperature sensitivity. This system is based on a pupil-plane aperture architecture used in a first generation system, but also includes advances in technology which improve system performance and utility. These include a flat panel dielectric antenna and W-band amplifiers, detectors, and processors. This second generation system will serve as the basis for a production millimeter-wave imager.

The design and testing of a close range Passive Millimeter Wave (PMMW) scanning thermal imager is described. While close range PMMW imaging has previously been applied to concealed weapon detection at ranges of a few meters, the imager under development here is designed to focus on targets at a range of a few tens of centimeters. In particular, the main design aim is to produce high resolution thermal maps suitable for medical imaging applications. Imaging at MMW frequencies offers greater- penetration depths in lossy dielectric media than conventional infrared (IR) imagers, although there is an obvious trade-off in spatial resolution.

Successful demonstration of a video rate 94 GHz camera for concealed weapons detection has lead to interest in commercialization of this system. Besides the basic physics of object detection, there are many details a practical system needs to consider for the transition from experiment to product. We describe improvements in the RF modules, back end electronics, and user interface, along with additional video images captured with the improved system. RF module improvements include better low noise amplifiers and RF switches, along with a two-temperature calibration method resulting in a noticeable improvement in image quality. Analog electronics now have a more rapid offset correction, better stability, and better dynamic range. The user interface will now permit image fusion and other user- friendly features. Indoor experiments with active illumination are reported.

We describe the status of, and present the first imagery obtained using an active uncooled mm-wave microbolometer- based imaging system. The microbolometers consist of 2 micrometers X 6 micrometers X .02 micrometers Nb films, coupled to the incident field by annular slot antennas. The system frequency is centered at 94 GHz. Imagery has been obtained using a mechanically scanned primary mirror and single-pixel bolometer, as a prototype for a 120-element focal-plane array still under development. With our most recent version of the optical system, high quality images, with clear detail on a spatial scale of 1 cm, have been obtained with acquisition times of 70 s and a target range of 1.3 m.

Thermal radiation of complex man-made objects as well as of our natural environment contains fully polarimetric information in many cases. A quasioptical, imaging radiometer system was designed and built up at DLR for the measurement of the four Stokes vector components. A polarimetric calibration procedure was developed and verified. Selected measurements have been carried out which indicate new possible applications.

This paper presents our group's most recent passive millimeter-wave (MMW) measurements made using a 94-GHz Stokes-vector radiometer. Included are images and analyses of treeline data. These data were collected to investigate the possible use of passive MMW sensors to perform the helicopter collision avoidance task. The treeline data presented were collected in both the summer and winter. The results of the analysis show that in the winter the detection of the treeline can be straightforward because of an often-low horizon sky brightness temperature. The contrast between the tree branches and the horizon are seen in the data to be about 10 - 15 K. The summer case, however, shows a horizon sky-to-tree brightness temperature ratio of about 1. A simple statistical analysis of the summer image shows that the trees, in our case, can be distinguished from the horizon sky based upon the statistical parameters alone.

This paper describes scene simulation in passive millimeter wave imaging. The appearance of metal and plastic objects are modeled lying flat on earth, as viewed at grazing incidence (from a ground platform) and perpendicularly (from an air platform), using a passive millimeter wave imager. The assumptions and essential physics behind the simulation are reviewed. The simulations are made in the atmospheric window at 90 GHz. Experimental data taken at 35 GHz is presented for comparison. It is demonstrated that metal objects have generally low radiation temperatures, in relation to their environments. Plastics, on the other hand, can have higher or lower radiation temperatures than their backgrounds, dependent on the polarization the type of earth, its condition and the amount of water present. In the cases demonstrated in this paper, the simulations agree well with the experimental data.

This paper describes a reflectometer, which can operate at either 35 or 94 GHz. A broadband signal is produced by a modulated noise source in either vertical or horizontal polarization. After reflection by the sample under test a super-heterodyne receiver detects this signal. The noise source and the receiver are mounted on two opposed 0.5 m parabolic antennas. These antennas are supported on arms, which can be rotated by stepper motors under computer control. The computer also controls the data logging system, which consists of a lock-in amplifier and analogue to digital converter. This instrument allows reflectivity to be measured for incidence angles from 20 to 70 degree(s) down to a minimum of 0.05%, which is sufficient to measure diffuse reflection.

This paper reviews the mathematical image processing methods available for super-resolving passive millimeter wave (PMMW) images. PMMW imaging has a number of advantages over infra- red (IR) and visible imaging in being able to operate under adverse weather conditions making it useful for all weather surveillance. The main disadvantage, however, is the size of aperture required to obtain usable spatial resolution. A typical aperture size would be 1 m diameter for a system operating at 94GHz. This aperture may be reduced if super- resolution techniques are employed. To achieve super- resolution non-linear methods of restoration are required in order to generate missing high frequency information. For thee to be genuine high frequencies it is necessary to restore the image subject to constraints. These constraints should apply directly to the scene content rather than to properties of any noise also present. The merits of the available super-resolution techniques are discussed with reference to sharpening noisy PMMW images. Any increase in sharpness of an image frequently results in an increase in the noise present. This can detract from the ability of a human observer to recognize an object in the scene. This problem is discussed with reference to a recent model of human perception.

As a result of its relatively short wavelength coupled with relatively high penetration of many materials, millimeter- wave imaging provides a powerful tool for penetrating bad weather or the detecting of concealed articles. By using a passive approach such as that implemented here, it is possible to accomplish these tasks without generating any form of radiation that could be either be picked up by an enemy or raise health concerns. In this paper we will detail software and algorithm improvements that have resulted in significant enhancements to the quality of our imagery.

In applications of PMMW imaging such as real-time video, fast restorations are needed to keep up with the frame rate. FFT-based restoration provides a fast implementation, but it does so at the expense of assuming that the blurring and deblurring are based on circular convolution. Unfortunately, when the opposite sides of the image do not match up well in intensity, this assumption can create significant artifacts across the image. The mathematically correct way to avoid boundary artifacts is to model the pixels outside the measured image window as unknown values in the restored image. However, this approach destroys the structure that makes the use of the FFT possible, since the unknown image is no longer the same size as the measured image. Thus, the restoration methods available for this problem no longer have the computational efficiency of the FFT. We propose a new restoration method for the unknown boundary approach that can be implemented in a fast and flexible manner. We decompose the restoration into a sum of two independent restorations. One restoration yields an image that comes directly from a modified FFT-based approach. This image can be thought of as a type of FFT restoration containing the usual boundary artifacts. The other restoration involves a set of unknowns whose number equals that of the unknown boundary values. This restoration represents the artifact correction image. By summing the two, the artifacts are canceled. Because the second restoration has a significantly reduced set of unknowns, it can be calculated very efficiently even though no circular convolution structure exists.

Several image processing procedures have been used for the enhancement and detection of weapons concealed underneath clothing in millimeterwave data. Specifically, registration, fusion, tracking, enhancement, segmentation, and recognition procedures have been successfully tested. These procedures are reviewed in this paper along with examples of their application.

Computational complexity is a major impediment to the real- time implementation of image restoration and super- resolution algorithms. Although powerful restoration algorithms have been developed within the last few years utilizing sophisticated mathematical machinery (based on statistical optimization and convex set theory), these algorithms are typically iterative in nature and require enough number of iterations to be executed to achieve desired resolution gains in order to meaningfully perform detection and recognition tasks in practice. Additionally, recent technological breakthroughs have facilitated novel sensor designs (focal plane arrays, for instance) that make it possible to capture mega-pixel imagery data at video frame rates. A major challenge in the processing of these large format images is to complete the execution of the image processing steps within the frame capture times and to keep up with the output rate of the sensor so that all data captured by the sensor can be efficiently utilized. Consequently, development of novel methods that facilitate real-time implementation of image restoration and super- resolution algorithms is of significant practical interest and will be the primary focus of this paper. The key to designing computationally efficient processing schemes lies in strategically introducing appropriate pre-processing and post-processing steps together with the super-resolution iterations in order to tailor optimized overall processing sequences for imagery data of specific formats. Three distinct methods for tailoring a pre-processing filter and integrating it with the super-resolution processing steps will be outlined in this paper. These methods consist of a Region-of-Interest (ROI) extraction scheme, a background- detail separation procedure, and a scene-derived information extraction step for implementing a set-theoretic restoration of the image that is less demanding in computation compared to the super-resolution iterations. A quantitative evaluation of the performance of these algorithms for restoring and super-resolving tactical imagery data, including in particular PMMW images, will also be presented.

This paper describes passive radiovision systems in 8-mm and 3-mm ranges. It contains short description of the systems, data processing algorithms and points out problems specific to multi-ray systems. Image enhancement methods are briefly considered. Several examples of radio thermal images of natural objects are presented herein.

To develop a useful, low cost, passive millimeter wave imaging system through batch processing of an integrated 2D antenna-coupled microbolometer focal plane array. To achieve high array sensitivity through sampling and post detection signal processing techniques adapted to this type of array. A review will be given of the thermal bridge and bolometer combination that is being developed, and the rationale for choosing the materials making up the bridge. The bolometer material and the bias circuitry for low noise and low power consumption will be discussed. An antenna design is also being developed, and will be discussed. Each receiver must satisfy multiple requirements. It must have a large RF bandwidth, match the feed from the imaging system, and also allow for close packing on the focal plane. The overall receiver geometry must provide for a convenient vacuum environment for the thermal bridges along with a ready access of the ROIC to the bolometer signals. The approach taken to achieve this will be discussed. An overview will also be given of the array geometry, the sampling strategy, and the signal processing approach taken for maximizing thermal sensitivity of passive millimeter wave video on a dynamic platform.