This PDF file contains the front matter associated with SPIE Proceedings Volume 9093, including the Title Page, Copyright information, Table of Contents, Introduction (if any), and Conference Committee listing.

This paper investigates techniques for using low probability of intercept (LPI) modulation techniques for forming synthetic aperture radar (SAR) imagery. This analysis considers a specific waveform type based upon Frank codes in providing for the LPI capability via phase shift keying (PSK) modulation. A correlation receiver that is matched to the transmitted waveform is utilized to generate a set of SAR data. This analysis demonstrates the ability to form SAR images based upon simulated radar measurements collected by a notional radar sensor that has ability to transmit and receive Frank-coded waveforms and to form SAR images based upon the results of a correlation receiver. Spotlight-mode SAR images are generated using the Frank-coded waveforms and their properties are analyzed and discussed.

This paper describes a Fast Factorized Back-Projection (FFBP) formulation that includes a fully integrated autofocus algorithm, i.e. the Factorized Geometrical Autofocus (FGA) algorithm. The base-two factorization is executed in a horizontal plane, using a Merging (M) and a Range History Preserving (RHP) transform. Six parameters are adopted for each sub-aperture pair, i.e. to establish the geometry stage-by-stage via triangles in 3-dimensional space. If the parameters are derived from navigation data, the algorithm is used as a conventional processing chain. If the parameters on the other hand are varied from a certain factorization step and forward, the algorithm is used as a joint image formation and autofocus strategy. By regulating the geometry at multiple resolution levels, challenging defocusing effects, e.g. residual space-variant Range Cell Migration (RCM), can be corrected. The new formulation also serves another important purpose, i.e. as a parameter characterization scheme. By using the FGA algorithm and its inverse, relations between two arbitrary geometries can be studied, in consequence, this makes it feasible to analyze how errors in navigation data, and topography, affect image focus. The versatility of the factorization procedure is demonstrated successfully on simulated Synthetic Aperture Radar (SAR) data. This is achieved by introducing different GPS/IMU errors and Focus Target Plane (FTP) deviations prior to processing. The characterization scheme is then employed to evaluate the sensitivity, to determine at what step the autofocus function should be activated, and to decide the number of necessary parameters at each step. Resulting FGA images are also compared to a reference image (processed without errors and autofocus) and to a defocused image (processed without autofocus), i.e. to validate the novel approach further.

The 3-D Fractional Fourier Transformation (FrFT) has unique applicability to multi-pass and multiple receiver Synthetic Aperture Radar (SAR) scenarios which can collect radar returns to create volumetric reflectivity data. The 3-D FrFT can independently compress and image radar data in each dimension for a broad set of parameters. The 3-D FrFT can be applied at closer ranges and over more aperture sampling conditions than other imaging algorithms. The FrFT provides optimal processing matched to the quadratic signal content in SAR (i.e. the pulse chirp and the spherical wave-front across the aperture). The different parameters for 3-D linear, circular, and orbital SAR case are derived and specifi…c considerations such as squint and scene extent for each scenario are addressed. Example imaged volumes are presented for linear, circular and orbital cases. The imaged volume is sampled in the radar coordinate system and can be transformed to a target based coordinate system. Advantages of the FrFT which extend to the 3-D FrFT include its applicability to a wide variety of imaging condition (standoff range and aperture sub-sampling) as well as inherent phase preservation in the images formed. The FrFT closely matches the imaging process and thus is able to focus SAR images over a large variation in standoff ranges specifi…cally at close range. The FrFT is based on the relationship between time and frequency and thus can create an image from an under-sampled wave-front. This ability allows the length of the synthetic aperture to be increased for a fixed number of aperture samples.

Interferometric Synthetic Aperture Radar (IFSAR) uses the phase difference between two SAR images acquired at different positions to infer ground topography. Conventional IFSAR technique is based on wideband transmitted waveforms. As a result, the interferometric phase forms an iso-Doppler surface containing the height information. In this work, we present a novel interferometric SAR technique using ultra-narrowband continuous waveforms to infer ground topography. Due to high Doppler resolution of the transmitted waveforms, we refer to this technique as the Doppler-IFSAR. We form SAR images by backprojecting onto iso-Doppler contours. We present the interferometric phase model for Doppler-IFSAR and outline the relationship between the height and interferometric phase.

Due to its computational efficiency, the polar format algorithm (PFA) is considered by many to be the workhorse for airborne synthetic aperture radar (SAR) imaging. PFA is implemented in spatial Fourier space, also known as “K-space”, which is a convenient domain for understanding SAR performance metrics, sampling requirements, etc. In this paper the mathematics behind PFA are explained and computed examples are presented, both using simulated data, and experimental airborne radar data from the Air Force Research Laboratory (AFRL) Gotcha Challenge collect. In addition, a simple graphical method is described that can be used to model and predict wavefront curvature artifacts in PFA imagery, which are due to the limited validity of the underlying far-field approximation. The appendix includes Matlab code for computing SAR images using PFA.

We present an analysis of the positioning errors in Backprojection (BP)-based Synthetic Aperture Radar (SAR) images due to antenna trajectory errors for a monostatic SAR traversing a straight linear trajectory. Our analysis is developed using microlocal analysis, which can provide an explicit quantitative relationship between the trajectory error and the positioning error in BP-based SAR images. The analysis is applicable to arbitrary trajectory errors in the antenna and can be extended to arbitrary imaging geometries. We present numerical simulations to demonstrate our analysis.

Three dimensional scene reconstruction with synthetic aperture radar (SAR) is desirable for target recognition and improved scene interpretability. The vertical aperture, which is critical to reconstruct 3D SAR scenes, is almost always sparsely sampled due to practical limitations, which creates an underdetermined problem. This papers explores 3D scene reconstruction using a convex model-based approach. The approach developed is demonstrated on 3D scenes, but can be extended to SAR reconstruction of sparsely sampled signals in the spatial and, or, frequency domains. The model-based approach enables knowledge-aided image formation (KAIF) by incorporating spatial, aspect, and sparsity magnitude terms into the image reconstruction. The incorporation of these terms, which are based on prior scene knowledge, will demonstrate improved results compared to traditional image formation algorithms. The SAR image formation problem is formulated as a second order cone program (SOCP) and the results are demonstrated on 3D scenes using simulated data and data from the GOTCHA data collect.¹ The model-based results are contrasted against traditional backprojected images.

In this paper we present a new generalized family of change detection algorithms for SAR imagery that includes traditional non-coherent and coherent processing as special cases. The parameterized family of algorithms, referred to as partially coherent change detection (PCCD), allows the user to select the level of coherence desired in the change detection algorithm. This and other settings of the algorithm enable one to specify the types of changes that are significant, thereby reducing the number of false alarms due to insignificant changes—such as foliage motion. Algorithm settings may also be applied spatially in order to support spatially varying levels of coherence based on scene content. Examples from synthetic and measured imagery demonstrate the efficacy of the new family of algorithms.

This paper describes a correction technique using stereo SAR (Synthetic Aperture Radar) that improves the image quality of video SAR. The video SAR mode provides a persistent view of a scene centered at the Motion Compensation Point (MCP). The radar platform follows a circular flight path. The effect of height-dependent layover in video SAR imagery is that objects higher than the image plane appear translated towards the along-track direction in the image plane. A Digital Elevation Map (DEM) provides the height of the MCP. A height error in the MCP results in a SAR image that is not centered at the MCP. An objective is to form a sequence of SAR images centering the MCP for each image. This paper details a correlation method performed on a pair of SAR images in order to estimate the layover. Using multiple pairs of images taken from video SAR imagery improves the reliability of the layover estimate. Synthetic target imagery with targets at non-zero heights demonstrates the method. The paper presents 2011-2013 flight data collected by General Atomics Aeronautical Systems, Inc. (GA-ASI) implementing the video SAR mode with correction for layover. The flight data demonstrates good video quality with the MCP centered in each video frame.

Many imaging algorithms have been designed assuming the absence of multiple scattering. In the 2013 SPIE proceeding, we discussed an algorithm for removing high order scattering components from collected data. In this paper, our goal is to continue this work. First, we survey the current state of multiple scattering in SAR. Then, we revise our method and test it. Given an estimate of our target reflectivity, we compute the multi scattering effects in our target region for various frequencies. Furthermore, we propagate this energy through free space towards our antenna, and remove it from the collected data.

Based on the fundamental scattering mechanisms of facetized computer-aided design (CAD) models, we are able to define expected contributions (EC) to the radar signature. The net result of this analysis is the prediction of the salient aspects and contributing vehicle morphology based on the aspect. Although this approach does not provide the fidelity of an asymptotic electromagnetic (EM) simulation, it does provide very fast estimates of the unique scattering that can be consumed by a signature exploitation algorithm. The speed of this approach is particularly relevant when considering the high dimensionality of target configuration variability due to articulating parts which are computationally burdensome to predict. The key scattering phenomena considered in this work are the specular response from a single bounce interaction with surfaces and dihedral response formed between the ground plane and vehicle. Results of this analysis are demonstrated for a set of civilian target models.

A process for selecting a sparse subset of features that maximize discrimination between target classes is described in a Bayesian framework. Demonstrated on high range resolution radar (HRR) signature data, this has the effect of selecting the most informative range bins for a classification task. The sparse Bayesian classifier (SBC) model is directly compared against Fisher's linear discriminant analysis (LDA), showing a clear performance gain with the Bayesian framework using HRRs from the publicly available MSTAR data set. The discriminative power of the selected features from the SBC is shown to be particularly dominant over LDA when only a few features are selected or when there is a shift in training and testing data sets, as demonstrated by training on a specific target type and testing on a slightly different target type.

Synthetic Aperture Radar (SAR) has an important contribution in monitoring ships in the littoral regions. This stems from the substantial information that SAR images have which can facilitate the ships detection operation. Coastline images produced by SAR suffer from many deficiencies which arise from the presence of speckles and strong signals returned from land and rough sea. The first step in many ship detection systems is to mark and reject the land in SAR images (land masking). This is performed to reduce the number of false alarms that might be introduced if the land is processed by ship detector. In this paper, two powerful methods for land masking are introduced. One is based on mathematical morphology while the other is based on Lee-Jurkevich coastline detection and mean estimator algorithm. From experimental results, the proposed methods give promising results for both strongly marking the land area in SAR images and efficiently preserving the details of coastlines as well.

With all of the new remote sensing modalities available, with ever increasing capabilities, there is a constant desire to extend the current state of the art in physics-based feature extraction and to introduce new and innovative techniques that enable the exploitation within and across modalities, i.e., fusion. A key component of this process is finding the associated features from the various imaging modalities that provide key information in terms of exploitative fusion. Further, it is desired to have an automatic methodology for assessing the information in the features from the various imaging modalities, in the presence of uncertainty. In this paper we propose a novel approach for assessing, quantifying, and isolating the information in the features via a joint statistical modeling of the features with the Gaussian Copula framework. This framework allows for a very general modeling of distributions on each of the features while still modeling the conditional dependence between the features, and the final output is a relatively accurate estimate of the information-theoretic J-divergence metric, which is directly related to discriminability. A very useful aspect of this approach is that it can be used to assess which features are most informative, and what is the information content as a function of key uncertainties (e.g., geometry) and collection parameters (e.g., SNR and resolution). We show some results of applying the Gaussian Copula framework and estimating the J-Divergence on HRR data as generated from the AFRL public release data set known as the Backhoe Data Dome.

Computational methods for electromagnetic scattering prediction have been an invaluable tool to the radar signal exploitation community. Scattering prediction codes can provide simulated data of varied levels of fidelity at a fraction of the cost of measured data. Software based on physical optics theory is presently the tool of choice for generating high-frequency scattering data. Currently available codes have extensive capabilities but are usually restricted in their distribution or application due to government or proprietary concerns and due to platform specific software designs. The Raider Tracer software, described in this paper, is a MATLAB-based scattering prediction code that was developed for open distribution to the broader research community.

Simultaneously estimating position x and velocity v of moving targets using only the measured phase ' from single-channel SAR is impossible because the mapping from (x, v) to φis many-to-one. This paper defines classes of equivalent target motion and solves the GMTI problem up to membership in an equivalence class using single-channel SAR phase data. Definitions are presented for endo- and exo-clutter that are consistent with the equivalence classes, and it is shown that most target motion can be detected, i.e. the set of endo-clutter targets is very small. We exploit the sparsity of moving targets in the scene to develop an algorithm to resolve target motion up to membership in an equivalence class, and demonstrate the effectiveness of the proposed technique using simulated data.

Along-track synthetic aperture radar (SAR) systems can be used to remove the bright stationary clutter in order to more easily detect and track moving targets within the scene. In this work, we derive a Cramér Rao Lower Bound (CRLB) on the localization error of moving targets in these systems in the presence of correlated and spatially-varying noise. The CRLB is used to determine the minimum range/cross-range position and velocity estimation errors and is evaluated over radar and target parameters. Bounds were compared both over design parameters such as the number of antennas and the coherent processing interval, as well as non-design parameters such as signal quality and clutter coherence. Furthermore, the bias sensitivity of the CRLB to small estimator biases was analyzed.

In this paper, experimental results for UHF wideband SAR imaging of humans on an open field and inside a forest is presented. The results show ability to detect the humans and suggest possible ways to improve the results. In the experiment, single channel wideband SAR mode of the UHF UWB system LORA developed by Swedish Defence Research Agency (FOI). The wideband SAR mode used in the experiment was from 220 to 450 MHz, thus with a fractional bandwidth of 0.68. Three humans walking and one stationary were available in the scene with one of the walking humans in the forest. The signature of the human in the forest appeared on the field, due to azimuth shift from the positive range speed component. One human on the field and the one in the forest had approximately the same speed and walking direction. The signatures in the SAR image were compared as a function of integration time based on focusing using the average relative speed of these given by GPS logs. A signal processing gain was obtained for the human in forest until approximately 15 s and 35 s for the human on the field. This difference is likely explained by uneven terrain and trees in the way, causing a non-straight walking path.

Along-Track Interferometry (ATI) has been widely used for ground moving target indication (GMTI) in airborne synthetic aperture radar (SAR) systems. In ideal cases, the ATI phase obtained using two phase centers that are aligned in the along-track dimension yield clutter-only pixels with zero phase. However, the platform's motion may create a cross-track displacement between the two phase centers and in turn offset the phase centers' baseline from the along track dimension. This cross-track offset leads to non-zero phase for clutter-only pixels, necessitating calibration for accurate GMTI. This paper proposes a blind calibration method to correct the along-track baseline error in ATI-SAR systems. The success of the proposed method is shown on a set of measured data from the Gotcha sensor.

We describe techniques for improving ground moving target indication (GMTI) performance in multi-channel synthetic aperture radar (SAR) systems. Our approach employs a combination of moving reference processing (MRP) to compensate for defocus of moving target SAR responses and space-time adaptive processing (STAP) to mitigate the effects of strong clutter interference. Using simulated moving target and clutter returns, we demonstrate focusing of the target return using MRP, and discuss the effect of MRP on the clutter response. We also describe formation of adaptive degrees of freedom (DOFs) for STAP filtering of MRP processed data. For the simulated moving target in clutter example, we demonstrate improvement in the signal to interference plus noise (SINR) loss compared to more standard algorithm configurations. In addition to MRP and STAP, the use of tracker feedback, false alarm mitigation, and parameter estimation techniques are also described. A change detection approach for reducing false alarms from clutter discretes is outlined, and processing of a measured data coherent processing interval (CPI) from a continuously orbiting platform is described. The results demonstrate detection and geolocation of a high-value target under track. The endoclutter target is not clearly visible in single-channel SAR chips centered on the GMTI track prediction. Detections are compared to truth data before and after geolocation using measured angle of arrival (AOA).

Combined synthetic aperture radar (SAR) and ground moving target indication (GMTI) radar modes simultaneously generate SAR and GMTI products from the same radar data. This hybrid mode provides the benefit of combined imaging and moving target displays as well as improved target recognition. However, the differing system, antenna, and waveform requirements between SAR and GMTI modes make implementing the hybrid mode challenging. The Air Force Research Laboratory (AFRL) Gotcha radar has collected wide-bandwidth, multi-channel data that can be used for both SAR and GMTI applications. The spatial channels on the Gotcha array are sparsely separated, which causes spatial grating lobes during the digital beamforming process. Grating lobes have little impact on SAR, which typically uses a single spatial channel. However, grating lobes have a large impact on GMTI, where spatial channels are used to mitigate clutter and estimate the target angle of arrival (AOA). The AOA ambiguity has a significant impact in the Gotcha data, where detections from the sidelobes and skirts of the mainlobe wrap back into the main scene causing a significant number of false alarms. This paper presents a sub-banding method to disambiguate grating lobes in the GMTI processing. This method divides the wideband SAR data into multiple frequency sub-bands. Since each sub-band has a different center frequency, the grating lobes for each sub-band appear at different angles. The method uses this variation to disambiguate target returns and places them at the correct angle of arrival (AOA). Results are presented using AFRL Gotcha radar data.

In multi-channel SAR systems, the detection of movers is carried out using techniques based on the Wiener filter. An important problem is the estimation of the moving target parameters such as along-track velocity, across-track velocity, and azimuth displacement. In this paper, the parameter estimation problem is formulated as a novel Bayesian state estimation problem. The unscented Kalman filter is applied to the problem and it is noted that it leads to good estimates.

Along-track interferometry (ATI) is used to detect ground moving targets against a stationary background in synthetic aperture radar (SAR) imagery. In this paper, we present a novel approach to multi-channel ATI wherein clutter cancellation is applied to each pixel of the multiple SAR images, followed by a Fourier transform to estimate range rate (Doppler). Range rate estimates allow us to compensate for the cross-range offset of the target, thus geo-locating the targets. We then present a number of benefits to this approach.

In previous work, we presented GMTI detection and geo-location results from the AFRL Gotcha challenge data set, which was collected using a 3-channel, X-band, circular SAR system. These results were compared against GPS truth for a scripted vehicle target. The algorithm used for this analysis is known as ATI/DPCA, which is a hybrid of along-track interferometry (ATI) and the displaced phase center antenna (DPCA) technique. In the present paper the use of ATI/DPCA is extended in order to detect and geo-locate all observable moving targets in the Gotcha challenge data, including both the scripted movers and targets of opportunity. In addition, a computationally efficient SAR imaging technique is presented, appropriate for short integration times, which is used for computing an image of the scene of interest using the same pulses of data used for the GMTI processing. The GMTI detections are then overlaid on the SAR image to produce a simultaneous SAR/GMTI map.

We consider the application of KronPCA spatio-temporal modeling techniques^{1, 2} to the extraction of spatiotemporal features for video dismount classification. KronPCA performs a low-rank type of dimensionality reduction that is adapted to spatio-temporal data and is characterized by the T frame multiframe mean μ and covariance ∑ of p spatial features. For further regularization and improved inverse estimation, we also use the diagonally corrected KronPCA shrinkage methods we presented in.¹ We apply this very general method to the modeling of the multivariate temporal behavior of HOG features extracted from pedestrian bounding boxes in video, with gender classification in a challenging dataset chosen as a specific application. The learned covariances for each class are used to extract spatiotemporal features which are then classified, achieving competitive classification performance.