Search

In recent years, deep neural networks have shown to achieve state-of-the-art performance on several classification and prediction tasks. However, these networks demand undesirable lengthy training times coupled with high computational resources (memory, I/O, processing time). In this work, we explore semi-random deep neural networks to achieve near real-time training and less computational resource usage. Although many works enhance the underlying hardware for real-time training, this work focuses on algorithmic optimization. It is shown that random projection networks with additional skipped connectivity and randomly weighted layers can boost the overall network performance while enabling for real-time training. Additionally, a tensor-train decomposition technique is leveraged to further reduce the model complexity of these networks. Our investigation accomplishes the following: 1) Tensor-train decomposition decreases the complexity of random projection networks, 2) compression of the fully connected hidden layer leads to a minimum $\sim40\times$ decrease in memory size, and 3) training under random projection networks can be achieved in near-real time.

Artificial intelligence (AI) seeks to emulate human reasoning, but is still far from achieving such results for actionable sensing in complex situations. Instead of emulating human situation understanding, machines can amplify intelligence by accessing large amounts of data, filtering unimportant information, computing relevant context, and prioritizing results (for example, answers to human queries) to provide human–machine shared context. Intelligence support can come from many contextual sources that augment data reasoning through physical, environmental, and social knowledge. We propose a decisions-to-data multimodal sensor and action through contextual agents (human or machine) that seek, combine, and make sense of relevant data. Decisions-to-data combines AI computational capabilities with human reasoning to manage data collections, perform data fusion, and assess complex situations (that is, context reasoning). Five areas of AI developments for context-based AI that cover decisions-to-data include: (1) situation modeling (data at rest), (2) measurement control (data in motion), (3) statistical algorithms (data in collect), (4) software computing (data in transit), and (5) human–machine AI (data in use). A decisions-to-data example is presented of a command-guided swarm requiring contextual data analysis, systems-level design, and user interaction for effective and efficient multimodal sensing and action.

Over the past decades, there have been many approaches to synthetic aperture radar (SAR) automatic target recognition (ATR). ATR includes detection, classification, and identification of targets, scene, and context. Recently, the explosion of methods for deep learning has attracted numerous researchers to compare machine learning methods for SAR ATR. This paper reviews many approaches conducted for SAR recognition and discerns the most promising approaches. Using the Moving and Stationary Target Acquisition and Recognition (MSTAR) data set, there are comparative methods to evaluate the advances from the community. The paper reviews many of the available techniques recently published to determine the state of the art in emerging concepts.

We present advanced Deep Learning (DL) techniques for robust Synthetic Aperture Radar (SAR) automatic target recognition (ATR) in the presence of noise and signal phase errors. Our research focuses on ensuring robust performance of SAR ATR algorithms under noise and adversarial attacks. Robust DL-based SAR ATR is paramount in operational scenarios such as disaster relief, search and rescue, and highly accurate object classification for autonomous vehicles. Our contributions, as described in this paper, include algorithm development and implementation of an advanced deep learning technique known as adversarial training (AT) to mitigate the detrimental effects of sophisticated noise and phase errors. Our research demonstrated that 1) AT improves performance under extended operating conditions, in some cases improving up to 10% over models without AT. 2) The use of AT improves performance when sinusoidal or wideband phase noise is present, in some cases gaining 40% in accuracy that would be lost in the presence of noise. 3) We find the model architecture has significant impact on robustness, with more complex networks showing a greater improvement from AT. 4) The availability of multi-polarization data is always advantageous. To our knowledge no one has provided an extensive analysis of the impact of adversarial machine learning (ML) on SAR image classification. Thus, this paper serves as a comprehensive research revealing the impact of adversarial attack and how to mitigate it.

This research details a new approach to optimize neural network architectures for Synthetic Aperture Radar (SAR) object classification on neuromorphic (e.g., IBM’s TrueNorth) and embedded platforms. We developed an algorithm to reduce the run-time and the power consumption of Deep Neural Networks (DNNs) classifiers by reducing the DNN model size required for a given object classification task. Reducing the model size reduces the number of mathematical operations performed, and the memory required, enabling computation on low size, weight and power (SWaP) hardware. We will provide our approach and results on relevant SAR data. Our entirely new approach starts with a very small multi-class convolution neural network (CNN) and replaces the standard negative log likelihood loss function with a single-class log loss function. We then generate an ensemble of small models trained for an individual class by varying the training data using a k-fold cross-validation and augmentation. This is done for each class and the resulting ensembles classify objects by finding the maximum average probability across each ensemble of single-class classifiers. We demonstrate 91-99 percent classification accuracy on three different datasets with composite networks that require almost 10 times fewer mathematical operations than SqueezeNet (a reduced parameter CNN with AlexNet performance).

In recent years deep neural networks have shown great advances in image processing tasks. For modern datasets, these networks require long training times due to backpropagation, high amount of computational resources for weight updates, and memory intensive weight storage. Exploiting randomness during the training of deep neural networks can mitigate these concerns by reducing the computational costs without sacrificing network performance. However, a fully randomized network has limitations for real-time target classification as it leads to poor performance. Therefore we are motivated in using semi-random deep neural networks to exploit random fixed weights. In this paper, we demonstrate that semi-random deep neural networks can achieve near real-time training with comparable accuracies to conventional deep neural networks models. We find that these networks are enhanced by the usage of skip connections and train rapidly at the cost of dense memory usage. With greater memory resources available, these networks can train on larger datasets at a fraction of the training time costs. These semi-random deep neural network architectures open up an avenue for further research in utilizing random fixed weights in neural networks.

Among various parameters, large scene object detection and classification accuracy depends on image quality. In general, deep neural networks (DNN) are trained to achieve a desired recognition accuracy on a set of targets. However, DNNs become tuned to the training data used and may not generalize to new unseen data artifacts. Classification accuracy of a previously trained DNN is significantly reduced when classification is run on an image altered with additive noise. In this research, we propose image pre-processing to reduce the impact of noise induced low classification accuracy. Our approach consists of applying compressive sensing inspired pre-processing techniques to noisy images. We then compare the object recognition accuracy of a pretrained model on pre-processed noisy images and unprocessed noisy images. We will present our technical method, results, and analysis on relevant synthetic aperture radar data.

Automatic Target Recognition (ATR) seeks to improve upon techniques from signal processing, pattern recognition (PR), and information fusion. Currently, there is interest to extend traditional ATR methods by employing Artificial Intelligence (AI) and Machine Learning (ML). In support of current opportunities, the paper discusses a methodology entitled: Systems Experimentation efficiency effectives Evaluation Networks (SEeeEN). ATR differs from PR in that ATR is a system deployment leveraging pattern recognition (PR) in a networked environment for mission decision making, while PR/ML is a statistical representation of patterns for classification. ATR analysis has long been part of the COMPrehensive Assessment of Sensor Exploitation (COMPASE) Center utilizing measures of performance (e.g., efficiency) and measures of effectiveness (e.g., robustness) for ATR evaluation. The paper highlights available multimodal data sets for Automated ML Target Recognition (AMLTR).

The resurgence of interest in artificial intelligence (AI) stems from impressive deep learning (DL) performance such as hierarchical supervised training using a Convolutional Neural Network (CNN). Current DL methods should provide contextual reasoning, explainable results, and repeatable understanding that require evaluation methods. This paper discusses DL techniques using multimodal (or multisource) information that extend measures of performance (MOP). Examples of joint multi-modal learning include imagery and text, video and radar, and other common sensor types. Issues with joint multimodal learning challenge many current methods and care is needed to apply machine learning methods. Results from Deep Multimodal Image Fusion (DMIF) using Electro-optical and infrared data demonstrate performance modeling based on distance to better understand DL robustness and quality to provide situation awareness.

During the 2018 SPIE DSS conference, panelists were invited to highlight the trends and use of artificial intelligence and deep learning (AI/DL) for information fusion. This paper highlights the common issues presented from the panel discussion. The key issues include: leveraging AI/DL coordinated with information fusion for: (1) knowledge reasoning and reasoning, (2) information fusion enhancement, (3) object recognition and tracking, (4) data with models fusion, and (5) deep multimodal fusion cognition strategies to support the user.

Acknowledgements •DARPA MSTAR Program •DARPA TRACE PM (Dr. John Gorman) •DARPA RFMLS PM (Mr. Paul Tilghman) •IBM's TrueNorth Program (Dr. Dharmendra Modha) •AFRL Colleagues (Dr. Qing Wu, Mr. Edmund Zelnio, Dr. Vince Velten, Ms. Lori Westerkamp, Dr. Eric Branch, Dr. John Nehrbass) •AFRL Information Directorate Summer Internship Program •Mr. Nathan Inkawhich (Duke University) •Mr. Matthew Inkawhich (Duke University)

This paper presents the design and analysis of orthogonal, Doppler-tolerant waveforms for waveform agile radar (e.g. multiple-input multiple-output (MIMO) radar) applications. Previous work has given little consideration to the design of radar waveforms that remain orthogonal when they are received. Our research is focused on: 1) developing sets of waveforms that are orthogonal on both transmit and receive, and 2) ensuring that these waveforms are Doppler tolerant when properly processed. Our proposed solution achieves the above-mentioned goals by incorporating direct sequence spread spectrum (DSSS) coding techniques on linear frequency modulated (LFM) signals. We call this spread spectrum coded LFM (SSCL) signaling. Our transmitted LFM waveforms are rendered orthogonal with a unique spread spectrum (SS) code. At the receiver, the echo signal will be decoded using its spreading code. In this manner, transmitted orthogonal waveforms can be match filtered only with the intended received signals. From analytical expressions of the waveforms we have designed and from simulation results, we found that: 1) cross-ambiguity function (CAF) of two LFM SS coded (orthogonal) waveforms is small for all delays and Dopplers (i.e. transmit and receive signals satisfy near orthogonality constraint); 2) the length of the SS code determines the amount of interference suppression (i.e., complete orthogonal or near orthogonal of the received signal); 3) we can process the same received signal in two different ways; one method can provide LFM signal resolution and the other method can provide ultrahigh resolution; 4) biorthogonal codes can be used to reduce bandwidth when code length is large. Our proposed waveforms inherit multiple attributes (e.g. chirp diversity, code diversity, frequency diversity) of diverse waveforms.

To measure visual acuity (VA) outcomes, complication rates, and the social impact of cataract surgery in a cohort who underwent surgery as children in Bangladesh.Case series.A total of 471 of 850 children from 6 Bangladeshi districts who had been identified as cataract blind using key informants (KIs) between 2004 and 2009 during the Bangladesh Childhood Cataract Campaign (BCCC) together with all those children not included in the BCCC database but in the Child Sight Foundation (CSF) database who had been identified as cataract blind.The subjects and families were contacted again by KIs and transported to local examination centers, where parents and subjects were administered a questionnaire and subjects underwent full ocular examination. Where operative data were available (15%), they were analyzed in conjunction with questionnaire and examination findings. Statistical analysis was performed using SPSS Statistics (IBM, Armonk, NY).Presenting and best-corrected visual acuities (BCVAs), cause(s) of poor outcome, postoperative refraction, and school attendance.A total of 407 of the participants had undergone bilateral surgery as children, with a mean follow-up of 8.8 years. The mean age at examination was 16 years (range, 5-28 years; standard deviation [SD], 4.6 years); 63% of those examined were male; 22% had a binocular presenting VA of20/60; and 53% were severely visually impaired or blind (VA20/200). After refraction, 33% had VA20/60 in their better eye and 33% had VA20/200. Factors that predicted poor VA in multivariate logistic regression analysis were nystagmus (P0.001), longer delay in presentation (P0.001), and magnitude of absolute spherical equivalent refractive error (P0.001). Some 50% had nystagmus, and 69% of those currently aged ≤16 years were attending school. Better acuity was associated with school attendance (P0.001), whereas gender was not.Approximately one third of all participants had a BCVA of ≥20/60 in their better eye. Amblyopia and nystagmus limited visual outcome, indicating the need for earlier detection and treatment. This is the first study to show the link between pediatric cataract outcome and access to education, a millennium development goal.

Many research efforts have been devoted to applying machine learning (ML) algorithms to the task of Automatic Target Recognition (ATR). In the 90’s, ML techniques such as Neural Networks were less popular due to various technological barriers and applications. Computational resources were scarce and expensive. Today, computational resources are not as expensive as in the past; however, an abundance of sensors and business data need to be analyzed in real-time. High performance computing (HPC) enables ML-based decision making in real-time or near real-time. This research explores the application of deep learning algorithms, specifically convolutional neural networks, to the task of ATR in synthetic aperture radar (SAR) imagery. We developed a Convolution Neural Networks (CNN) architecture for achieving ATR in SAR imagery and found that classification accuracy levels of 99% can be achieved through the application of neural networks. We used graphics processing units (GPU) to accomplish the computational tasks.

Wideband radar waveforms that employ spread-spectrum techniques were investigated and experimentally tested. The waveforms combine bi-phase coding with a traditional LFM chirp and are applicable to joint SAR-GMTI processing. After de-spreading, the received signals can be processed to support simultaneous GMTI and high resolution SAR imaging missions by airborne radars. The spread spectrum coding techniques can provide nearly orthogonal waveforms and offer enhanced operations in some environments by distributing the transmitted energy over a large instantaneous bandwidth. The LFM component offers the desired Doppler tolerance. In this paper, the waveforms are formulated and a shift-register approach for de-spreading the received signals is described. Hardware loop-back testing has shown the feasibility of using these waveforms in experimental radar test bed.

There is a history and understanding of exploiting moving targets within ground moving target indicator (GMTI) data, including methods for modeling performance. However, many assumptions valid for GMTI processing are invalid for synthetic aperture radar (SAR) data. For example, traditional GMTI processing assumes targets are exo-clutter and a system that uses a GMTI waveform, i.e. low bandwidth (BW) and low pulse repetition frequency (PRF). Conversely, SAR imagery is typically formed to focus data at zero Doppler and requires high BW and high PRF. Therefore, many of the techniques used in performance estimation of GMTI systems are not valid for SAR data. However, as demonstrated by papers in the recent literature, 1-11 there is interest in exploiting moving targets within SAR data. The techniques employed vary widely, including filter banks to form images at multiple Dopplers, performing smear detection, and attempting to address the issue through waveform design. The above work validates the need for moving target exploitation in SAR data, but it does not represent a theory allowing for the prediction or bounding of performance. This work develops an approach to estimate and/or bound performance for moving target exploitation specific to SAR data. Synthetic SAR data is generated across a range of sensor, environment, and target parameters to test the exploitation algorithms under specific conditions. This provides a design tool allowing radar systems to be tuned for specific moving target exploitation applications. In summary, we derive a set of rules that bound the performance of specific moving target exploitation algorithms under variable operating conditions.

Advanced spread spectrum linear frequency modulated (LFM) waveforms are being developed for advanced capability synthetic aperture radar (SAR) and ground moving target indication (GMTI) applications. We have demonstrated by analysis and simulation the feasibility of these new type waveforms and are now in the process of implementing them in hardware. The basic approach is to combine a traditional LFM radar waveform with a direct sequence spread spectrum (DSSS) waveform, and then on receive to de-spread the return and capture the resultant LFM return for traditional matched filter processing and enhanced SAR and GMTI. We show the analysis, simulation and some preliminary hardware results.

BAE Systems has developed a baseline real-time selected vehicle (SV) radar tracking capability that successfully tracked multiple civilian vehicles in real-world traffic conditions within challenging semi-urban clutter. This real-time tracking capability was demonstrated in laboratory setting. Recent enhancements to the baseline capability include multiple detection modes, improvements to the system-level design, and a wide-area tracking mode. The multiple detection modes support two tracking regimes; wide-area and localized selected vehicle tracking. These two tracking regimes have distinct challenges that may be suited to different trackers. Incorporation of a wide-area tracking mode provides both situational awareness and the potential for enhancing SV track initiation. Improvements to the system-level design simplify the integration of multiple detection modes and more realistic SV track initiation capabilities. Improvements are designed to contribute to a comprehensive tracking capability that exploits a continuous stare paradigm. In this paper, focus will be on the challenges, design considerations, and integration of selected vehicle tracking.

Synthetic aperture radar (SAR) imaging is often used to image an area using airborne platforms that generate a large aperture by virtue of the platform motion. Large apertures generate a large synthetic array providing fine cross-range resolution, and together with wide bandwidth waveforms that provide high range resolution, fine resolution images can be generated. SAR algorithms make use of coherent phase compensation from various pulses for focusing and the technique works exceedingly well for scenes containing stationary scattering centers. When moving targets are present, their images are smeared and shifted due to the motion, and to take advantage of this shift, nearby receiver plates are used to form multiple SAR images and together with along track interferometry (ATI), it generates a phase factor that can be used to detect moving target presence. This paper examines the distribution of the phase variable used in ATI for zero mean Complex Gaussian clutter/target data, and uses the results to address the target in clutter problem as a hypothesis testing problem to compute the probability of detection/false alarm as a function of target to clutter ratio and its velocity.

This paper provides analytical principles to relate the signature of a moving target to parameters in a SAR system. Our objective is to establish analytical tools that could predict the shift and smearing of a moving target in a subaperture SAR image. Hence, a user could identify the system parameters such as the coherent processing interval for a subaperture that is suitable to localize the signature of a moving target for detection, tracking and geolocating the moving target. The paper begins by outlining two well-known SAR data collection methods to detect moving targets. One uses a scanning beam in the azimuth domain with a relatively high PRF to separate the moving targets and the stationary background (clutter); this is also known as Doppler Beam Sharpening. The other scheme uses two receivers along the track to null the clutter and, thus, provide GMTI. We also present results on implementing our SAR-GMTI analytical principles for the anticipated shift and smearing of a moving target in a simulated code. The code would provide a tool for the user to change the SAR system and moving target parameters, and predict the properties of a moving target signature in a subaperture SAR image for a scene that is composed of both stationary and moving targets. Hence, the SAR simulation and imaging code could be used to demonstrate the validity and accuracy of the above analytical principles to predict the properties of a moving target signature in a subaperture SAR image.

In a recent report, it has been brought to notice that close to five million people in the USA are affected by chronic wounds and billions are spent annually for their treatment. Despite advances in chronic wound management over the past decades, many patients afflicted with chronic wounds fail to heal or their ulcers recur. Why is this so?

In this research, we introduce a signal processing framework for joint GMTI and SAR algorithms that is based on orthogonal (transmit and receive) waveforms. Traditionally, radar systems are configured to operate either in GMTI or SAR processing mode, but not both simultaneously. This is due to the fact that operational parameters for these two modes are quite different. For example, exoclutter GMTI processing requires a high pulse repetition frequency (PRF), but a high PRF results in increased range ambiguity - and an increased processing burden - in SAR imaging. We propose combining diverse, orthogonal waveforms and introducing corresponding processing techniques to reduce the problems and complexities of joint GMTI and SAR exploitation. For the exoclutter GMTI problem, the necessary high-PRF pulse train will be used to achieve finer Doppler resolution for detecting fast moving objects. For the endoclutter GMTI and SAR imaging problem, we will transmit low PRF pulses. The goal for low PRF pulses for endoclutter GMTI and SAR imaging is to ensure that range ambiguity issue has been addressed. These new approaches will achieve following benefits: (1) accomplish GMTI and SAR processing concurrently by eliminating the complexities associated with reconfiguring a radar system, (2) more efficiently use bandwidth by employing appropriate bandwidth for exoclutter GMTI pulses and SAR image formation pulses, and (3) reduce range ambiguity issue associated with high PRF operation.

40-years of digital SAR and Slow GMTI technology is traced from the first system to demonstrate real-time digitally correlated SAR from an intentionally maneuvering platform [1] to the current lite-weight dual-channel radar (DCR) providing simultaneous SAR and GMTI data. The Dual-Channel Radar (DCR) has been developed providing lite-weight SAR GMTI capability for Small UAVs. The prototype radar weighs 5-lbs and has demonstrated the extraction of ground moving targets (GMTs) embedded in highresolution SAR imagery data. Sum and difference channel data is used in a DPCA algorithm to extract the GMTs and display them on the Sum channel high resolution SAR image. Heretofore this type of capability has been reserved for much larger systems such as the JSTARS. Previously small liteweight SARs featured only a single channel and only displayed SAR imagery. With the advent of this new capability, SAR GMTI performance is now possible for small UAV class radars for DoD and DHS applications. The DCR is the culmination of multiple Phase II and Phase II plus SBIR efforts over the past 10-years, since 2002, for the Army, DARPA and AFRL.

A small and lightweight dual-channel radar has been developed for SAR data collections. Using standard Displaced Phase Center Antenna (DPCA) radar digital signal processing, SAR GMTI images have been obtained. The prototype radar weighs 5-lbs and has demonstrated the extraction of ground moving targets (GMTs) embedded in high-resolution SAR imagery data. Heretofore this type of capability has been reserved for much larger systems such as the JSTARS. Previously, small lightweight SARs featured only a single channel and only displayed SAR imagery. Now, with the advent of this new capability, SAR GMTI performance is now possible for small UAV class radars.

In this paper, we present an approach to the design of orthogonal, Doppler tolerant waveforms for diversity waveform radar (e.g. MIMO radar). Previous work has given little consideration to the design of radar waveforms that remain orthogonal when they are received. Our research is focused on: (1) developing sets of waveforms that are orthogonal on both transmit and receive, and (2) ensuring that these waveforms are Doppler tolerant when properly processed. Our proposed solution achieves the above mentioned goals by incorporating direct sequence spread spectrum (DSSS) coding techniques on linear frequency modulated (LFM) signals. We call it Spread Spectrum Coded LFM (SSCL) signaling. Our transmitted LFM waveforms are rendered orthogonal with a unique spread spectrum code. At the receiver, the echo signal will be decoded using it's spreading code. In this manner, transmitted orthogonal waveforms can be match filtered only with the intended received signals. From analytical expressions of the waveforms we have designed and from simulation results, we found that: (a) cross-ambiguity function of two LFM spread spectrum coded (orthogonal) waveforms is small for all delays and Dopplers (i.e. transmit and receive signals satisfy the orthogonality constraint), (b) The length and type of the spread spectrum code determines amount of suppression (i.e. complete orthogonal or near orthogonal of the received signal), (c) We can process the same received signal in two different ways; one method can provide LFM signal resolution and the other method can provide ultra high resolution.

A Dual-Channel Radar (DCR) has been developed providing lite-weight SAR GMTI capability for Small UAVs. The prototype radar weighs 5-lbs and has demonstrated the extraction of ground moving targets (GMTs) embedded in high-resolution SAR imagery data. Sum and difference channel data is used in a DPCA algorithm to extract the GMTs and display them on the Sum channel high resolution SAR image. Heretofore this type of capability has been reserved for much larger systems such as the JSTARS. Previously small liteweight SARs featured only a single channel and only displayed SAR imagery. Now, with the advent of this new capability, SAR GMTI performance is now possible for small UAV class radars.

This paper presents a design for the parallel processing of synthetic aperture radar data using one or more Field Programmable Gate Arrays (FPGAs). Our design supports real-time computation of a two-dimensional image from a matrix of echo pulses and their corresponding response values. Components of this design include: (a) central processing pipeline to perform back projection calculations, (b) pre-fetch cache to minimize external memory access latency, (c) memory bridge that serves as the primary on-chip storage for pulse data, and (d) a pixel queue to direct image data in and out of the pipeline. Design parameters may be adjusted to achieve optimum performance, and multiple instances of this design may be replicated on-chip to achieve prespecified performance objectives. We provide a complexity analysis as a function of the input and output parameters. Simulation results based on an implementation of this design show that our design achieves 160 GFLOPs per instance on a simulated Altera Stratix III EP3SL150 FPGA, and scales well for output image size ranging from 500 × 500 pixels to 5,000 × 5,000 pixels.

Traditionally, distinct radar modes have been employed to accomplish specific tasks such as imaging an area of interest, or detecting and tracking moving targets. Staring circular synthetic aperture radar (S-CSAR) provides unique opportunities for exploitation of radio frequency (RF) data collected over a large ground spot. The same phase history may be processed in different manners to generate simultaneous S-CSAR products such as 2-D Video SAR, coherent and non-coherent change detection (CCD and NCD), and ground moving target indication (GMTI). Advanced signal processing techniques can take advantage of the S-CSAR geometry to produce 3-D scene reconstructions. The ability to transmit, record and process large volumes of S-CSAR data, to create high fidelity exploitation products, in real-time, poses significant challenges. This paper addresses several open problems in this research area.

A major area of focus for the Air Force is sensor performance in urban environments. Aircraft with multiple sensor modalities, such as Synthetic Aperture RADAR (SAR), Infrared (IR), and Electro-Optics (EO), are essential for intelligence, surveillance, and reconnaissance (ISR) of current and future urban battlefields. Although applications exist for visualization of these types of imagery, they usually require at least a laptop computer and internet connection. Field operatives need to be able to access georeferenced information about imagery as part of a Geographic Information System (GIS) on mobile devices. The iPod/iPhone has a 640x480 resolution multi-touch display, making it an excellent device for interacting with georeferenced imagery. We created an iPhone application that loads SAR imagery and allows the user to interact with it. The user multi-touch interface provides pan and zoom capabilities as well as options to change parameters relating to the query. We describe how operatives in the field can use this application to investigate SAR and GIS related problems on the iPhone mobile device, which otherwise would require a computer and Internet connection.

This document describes a challenge problem whose scope is two-fold. The first aspect is to develop SAR CCD algorithms that are applicable for X-band SAR imagery collected in an urban environment. The second aspect relates to effective data compression of these complex SAR images, where quality SAR CCD is the metric of performance. A set of X-band SAR imagery is being provided to support this development. To focus research onto specific areas of interest to AFRL, a number of challenge problems are defined. The data provided is complex SAR imagery from an AFRL airborne X-band SAR sensor. Some key features of this data set are: 10 repeat passes, single phase center, and single polarization (HH). In the scene observed, there are multiple buildings, vehicles, and trees. Note that the imagery has been coherently aligned to a single reference.

Three-dimensional (3-D) spotlight-mode synthetic aperture radar (SAR) images of point scatterers provide insight into the achievable effectiveness of exploitation algorithms given a variety of operating parameters such as elevation angle, azimuth or synthetic aperture extent, and frequency bandwidth. Circular SAR, using 360 degrees of azimuth, offers the benefit of persistent surveillance and the potential for 3-D image reconstruction improvement compared with limited aperture SAR due in part to the increase in favorable viewing angles of unknown objects. The response of a point scatter at the origin, or center of the imaging scene, is known and has been quantified for circular SAR in prior literature by a closed-form solution. The behavior of a point scatterer radially displaced from the origin has been previously characterized for circular SAR through implementation of backprojection image reconstructions. Here, we derive a closed-form expression for the response of an arbitrarily located point scatterer given a circular flight path. In addition, the behavior of the response of an off-center point target is compared to that of a point scatterer at the origin. Symmetries within the 3-D point spread functions (PSFs), or impulse response functions (IPRs), are also noted to provide knowledge of the minimum subset of SAR images required to fully characterize the response of a particular point scatterer. Understanding of simple scattering behavior can provide insight into the response of more complex targets, given that complicated targets may sometimes be modeled as an arrangement of geometrically simple scattering objects.