The U-2 is the premier high altitude Armed Services platform for all reconnaissance and intelligence data collection. The U-2's primary imaging sensor is the ASARS-2 which is an all weather day/night radar capable of providing near real-time imagery through various collection modes. The Air Force Reconnaissance System Program Office through the Defense Airborne Reconnaissance Office and Secretary of the Air Force, Directorate of Information Dominance is currently upgrading the airborne radar, data-links, and round processing and exploitation components to provide increased coverage, improved image quality, and provide a near real- time targeting capability for precision guided munitions. The ASARS-2 Improvement Program (AIP) will be discussed to include the hardware changes to the sensor, software changes to the ground station and new exploitation tools. AIP will increase the area coverage of the current ASARS-2 system through the use of new airborne components. The system will employ full on-board processing allowing detected, complex, or video phase history products and maintain the requirements for the collected sensor data to fit within the current 274 Mb/sec bandwidth of the air-ground Data Link. Through the use of complex imagery data, the AIP program will augment the ground processing and exploitation segment with specialized imagery data, the AIP program will augment the ground processing and exploitation segment with specialized imagery enhancing tools. Some of the tools under evaluation include; spatially variant apodization, interferometric SAR, multi-view, coherent change detection, super-resolution, and moving target indication products. AIP will also significantly improve the geolocation targeting accuracy of SAR imagery for use with GPS aided munitions.

Recent advances in the areas of phase history processing, interferometry, and radargrammetric adjustment have made possible extremely accurate data extraction from synthetic aperture radar (SAR) imagery data. The potential gain from interferometric exploitation is significant since accuracy of measurements can theoretically be determined to within a resolution element of wavelength dimension. Recent work by the authors has shown that the main barrier to accurate and efficient elevation extraction is the measurement jitter caused by terrain variations, which overlay differently in the two SAR images. A unique combination of advanced photogrammetric and signal processing techniques is described which makes possible more accurate extraction of metric position and elevation model data form multiple pass SAR. This paper addresses the accuracy achievable from repeat passes of the ERS-1 and SIR-C spaceborne platforms and is also applicable to airborne SAR platforms.

This paper studies the problem of extracting target features via synthetic aperture radar (SAR) in the presence of uncompensated aperture motion errors. A parametric data model for a spotlight-mode SAR system is established. The Cramer-Rao bounds (CRBs) for the parameters of the data model are also derived. The CRB analysis shows that the unknown motion errors can significantly affect the accuracy of a common shift of the scatterer positions in the cross- range direction, but have little effect on other target parameters including the accuracy of the relative positions in the cross-range direction. A relaxation-based MCRELAX algorithm for estimating both target features and motion errors is devised. Simulation results show that he mean- squared errors of the parameter estimates obtained by using the MCRELAX algorithm can approach the corresponding CRBs.

Wavefront curvature defocus effects can occur in spotlight- mode SAR imagery when reconstructed via the well-known polar formatting algorithm under certain scenarios that include imaging at close range, use of very low center frequency, and/or imaging of very large scenes. The range migration algorithm, also known as seismic migration, was developed to accommodate these wavefront curvature effects. However, the along-track upsampling of the phase history data required of the original version of range migration can in certain instances represent a major computational burden. A more recent version of migration processing, the frequency domain replication and downsampling (FReD) algorithm, obviates the need to upsample, and is accordingly more efficient. In this paper we demonstrate that the combination of traditional polar formatting with appropriate space-variant post- filtering for refocus can be as efficient or even more efficient than FReD under some imaging conditions, as demonstrated by the computer-simulated results in this paper. The post-filter can be pre-calculated from a theoretical derivation of the curvature effect. The conclusion is that the new polar formatting with post filtering algorithm should be considered as a viable candidate for a spotight-mode image formation processor when curvature effects are present.

Conventional radar range-Doppler imaging uses the Fourier transform to retrieve Doppler information. Due to a target's complex motion, the Doppler frequency shifts are actually time-varying. By using the Fourier transform, the Doppler spectrum becomes smeared and the image becomes blurred. In this paper, we propose a time-frequency based ISAR image formation technique to replace the Fourier based ISAR image formation for resolving image blurring problem without resorting to sophisticated motion compensation algorithms. Some high resolution time-frequency transforms are discussed and examples of applying the time-frequency based image formation to simulated ISAR data are given.

In SAR/ISAR imaging, estimation of motion parameters of moving objects is needed in order to compensate for translational motion that causes image blurring. Phase parameter estimation from signals corresponding to one scatterer buried in complex white Gaussian noise is a problem for which various successful techniques have been devised. For multicomponent signals, it is difficult to use these parametric methods. Isolating the signal's components in order to do the estimation can help. The authors have previously proposed the isolation of scatterers for instantaneous frequency estimation. This process allows not only the extraction of components but also the manipulation of each object separately. In this paper, we study the use of time-frequency filtering on parameter estimation for a complex exponential signal with polynomial phase modulation. Our study is limited to the use of the Discrete Polynomial- Phase Transform (DPT) of Peleg and Friedlander. Our approach is to simplify the problem of parameter estimation from multicomponent signals by isolating them and then performing the estimation on each one separately. The process of isolating components is also shown to be improved by applying superresolution in the computation of the time- frequency representation. We also formulate an approach using the DPT for motion parameter estimation on a signal model for stepped-frequency ISAR returns from a scatterer.

A popular approach to SAR ATR is to use a hierarchical arrangement of stages, aimed at narrowing the field of interest and concentrating detection resources in those areas with the highest probability of containing a target. CFAR detectors have played a prominent role in such systems. So have morphological processing techniques, but to a lesser extent. The work reported in this paper follows the traditional hierarchical approach and employs CFAR and morphological principles. With the goal of improving detection accuracy and reducing the false alarm rate, we have explored and developed some new variations on the theme. The baseline CFAR and morphology algorithms have been significantly modified in proposed techniques to better match SAR target signatures. In addition, the modified algorithms have the ability to locate targets in a fast and efficient manner, allowing for the possibility of real-time implementation. Experimental evaluations indicate that both approaches perform well, with further performance gains noted when the algorithms are merged. The advantages and drawbacks of each algorithm are presented and experimental results are shown for a variety of test cases.

This paper summarizes recent results in several areas of SAR target detection. We present detection algorithms designed for parallel computation and a multiresolution segmentation algorithm used to extract the context of target detections and adapt the detection algorithms to the regions being processed. We consider cell averaging and order statistic constant false alarm rate (CFAR) detectors with a Weibull clutter background on SIMD linear arrays with simple fixed point arithmetic units.After considering the parallel computation of the various CFAR algorithms, we focus on the problem of segmenting the SAR imagery to obtain the context of the detections and to adapt the detector parameters for the area under test. Our multiresolution segmentation uses maximum likelihood estimation possibly followed by the iterative conditional modes algorithm at each resolution with merging accomplished using confidence levels. The algorithms are tested on 1 ft resolution SAR intensity images taken from the TESAR sensor onboard the Predator unmanned aerial vehicle.

This paper studies feature extraction of targets consisting of both trihedral and dihedral corner reflectors via synthetic aperture radar. A mixed data model is introduced to describe the target features. The Cramer-Rao bounds (CRBs) for the parameter estimates of the data model are also derived. Two algorithms, the FFTB algorithm and the NLS algorithm, are devised to estimate the model parameters. Numerical examples show that the parameter estimates obtained with both algorithms reach the CRBs as the SNR increases. The parameter estimates obtained with the NLS algorithm start to achieve the CRB at a lower SNR than those with the FFTB algorithm, while the latter algorithm is computationally more efficient.

A physical optics formalism is used to establish a first- principles analysis for discriminating specular returns from diffuse returns in a 1D synthetic aperture radar. The optimum Neyman-Pearson detection processor is shown to substantially outperform the conventional, full-resolution SAR images for extended specular targets embedded in diffuse clutter plus receiver noise. Significant performance advantages, relative to conventional full-resolution SAR processing, are also shown to accrue for optimum Neyman- Pearson reception of an extended diffuse target embedded solely in receiver noise.

Recent developments in optimal trade-off based composite correlation filter methods have improved the recognition and classification of an object over a range of image distortions. We extend the capability of the distance classifier correlation filter introduced by Mahalanobis et al by using he optimal trade-off between different correlation criteria. These correlation filters can be used for the automatic target cueing or recognition of synthetic aperture radar (SAR) images. In this paper we will present results of designing these distortion-tolerant filters with simulated SAR imagery and testing with simulated SAR target images inserted into real SAR backgrounds.

The feature space trajectory representation and neural network is used for classification and pose estimation of distorted objects in SAR data. New feature spaces and techniques to extend the concept to multiple classes are emphasized with initial four class results. On 4 class data, we obtain P_c equals 98.3 percent and clutter P_FA equals 0.026/km².

New variations of MINACE filters are considered for detection and recognition of objects and rejection of clutter in synthetic aperture radar (SAR) data. Multiple classes of objects with distortions present must be handled. Good performance in the presence of object obscurations is also needed. We present initial very attractive results that aid different modules in a SAR automatic target recognition processor. For 4 class test set data, we show detection performance of P_D > 99.8 percent with P_FA is congruent to 0.026/km²; for classification we show P_C > 96.6 percent with P_FA is congruent to 0.026/km². Selected tests with noise and obscurations present and of object in real SAR background are also included. All results used full shift-invariance tests.

The paper presents a methodology and GETP experimental system for rapid SAR target signature generation from limited initial sensory data. The methodology exploits and integrates the following four processes: (1) analysis of initial SAR image signatures and their transformation into higher-level blob representation, (2) blob modeling, (3) genetic inheritance modeling to generate new instances of a target model in blob representation, and (4) synthesis of new SAR signatures from genetically evolved blob data. The GETP system takes several SAR signatures of the target and transforms each signature into more general scattered blob graphs, where each blob represents local energy cluster. A single graph node is describe by blob relative position, confidence, and iconic data. Graph data is forwarded to the genetic modeling process while blob image is stored in a catalog. Genetic inheritance is applied to the initial population of graph data. New graph models of the target are generated and evaluated. Selected graph variations are forwarded to the synthesis process. The synthesis process restores target signature from a given graph and a catalog of blobs. The background is synthesized to complement the signature. Initial experimental results are illustrated with 64 X 32 image sections of a tank.

In this paper,we present an initial development of simple approximations for the performance of real-aperture radar ATR systems. In particular, we develop estimates of target separability using channel-capacity-like approximations, based on sensor constraints, and combinatoric-driven approximations, based on constraints imposed by the target. We then use these approximations to form estimates of the classification error probability for the ATR system. Finally, these estimates are compared with classification results for a small data set containing both synthetic and measured down-range responses form aircraft objects.

Our work has focused on deformable template representations of geometric variability in automatic target recognition (ATR). Within this framework we have proposed the generation of conditional mean estimates of pose of ground-based targets remotely sensed via forward-looking IR radar (FLIR) systems. Using the rotation group parameterization of the orientation space and a Bayesian estimation framework, conditional mean estimators are defined on the rotation group with minimum mean squared error (MMSE) performance bounds calculated following. This paper focuses on the accommodation of thermodynamic variation. Our new approach relaxes assumptions of the target's underlying thermodynamic state, expanding thermodynamic state as a scalar field. Estimation within the deformable template setting poses geometric and thermodynamic variation as a joint inference. MMSE pose estimators for geometric variation are derived, demonstrating the 'cost' of accommodating thermodynamic variability. Performance is quantitatively examined, and simulations are presented.

We show how boundary methods, a collection of tools used for distribution analysis, can be used to rank order ATR feature sets. Using boundary methods, we can establish which feature set is most appropriate for classification. We demonstrate experimentally that the derived ranking is consistent with alternative ranking techniques based on Bayes error. However, the use of boundary methods does not require extensive Monte-Carlo simulations and is not predicted on the assumption that the data is normally distributed.

This paper explores using linear regression and artificial neural networks (ANN) to model the performance of an ATR algorithm based on a given set of data. Here, a probability of detection response surfaces as a function of relevant parameters is simulated. It is then shown that this surface can be approximated using either linear regression or an ANN with good results. These regression surfaces can provide valuable information to the ATR developer/customer in terms of trying to predict ATR performance in untested areas. The application of this ATR performance modeling methodology becomes clear when we consider applying it to a common problem, such as air-to-ground target detection, where the changing parameters of the target can give a good set of data points from which to build the response curve.

Camouflage is an attempt to obscure the signature of a target and also to match its background. The goal is to make the detection and recognition of the target more difficult, whether the observer is a man, a machine, or both. In this paper, we are concerned specifically with the task of target discrimination. In this context, the signature strength of a target in a sensed image is equivalent to the distinctness of the image pattern representing the target from the pattern of its specific background. We complete both first- order and second-order target signature metrics for a set of 15 test images which consist of a wide variety of naturally occurring target and background patterns. To compare the quantitative measures of target distinctness to human judgments of the same attribute, the sets images were used as stimuli in a paired comparison experiment with human observers. For each comparison, the observer was asked to choose which of the two images possesses a target pattern that is more distinct from its background. The raw judgments from 20 observers were used to estimate scale values for the stimuli, indicating relative amounts of perceived target distinctness in the images. Of the metrics considered, a second-order metric based on a model of image texture was most strongly correlated with the scale values. These results are most applicable to the areas of camouflage assessment and design.

Statistical communication theory is used to develop the structure and performance of quasi-optimal recognition processors for 3D coherent laser radar range imagery. Generalized likelihood-ratio tests and receiver operating characteristics are presented for detection and recognition scenarios involving a variety of unknown object and background parameters.

This paper introduces concepts that, we hope, will help move the discussion of ATR evaluation in a direction that addresses long standing difficulties associated with getting test results that are meaningful to the program managers as they compare performance across technologies, to the users as they consider applications, and to the developers as they consider alternative approaches to the many ATR challenges. The paper is motivated by the recent need to independently evaluate an ATR system whose design is model-driven, particularly the DARPA/WL moving and stationary target acquisition and recognition program. There are two complementary classes of concepts. One class, which we call performance, includes accuracy, extensibility, robustness, and utility. These performance concepts encourage explicit consideration of the relationship between the test data, the training data, and data from modeled conditions. The other class, which we call cost includes efficiency, scalability, and synthetic trainability. Cost concepts help bring out some of the unique characteristics of the costs associated with ATR design and operation.

This paper defines the ATR problem outside the boundaries of the statistical pattern recognition (SPR) problem. It is believed that the state of the art supports successful application of SPR strategies to solve recognition problems and to the extent that the automatic target recognition (ATR) problem and the SPR problem are the same, the ATR problem is quite solvable. However, ATR remains problematic is its full realization and promise and has only been solved under a set of constrained conditions - those which map into the SPR problem. These are problems where the conditions of the training set are totally representative of the conditions under the test set. The purpose of this paper is to facilitate further progress in ATR development by defining the ATR problem in a more general way that is believed to be more representative of the actual ATR problem facing various ATR users rather than the more restricted SPR definition.

We consider the problem of detecting anisotropic scattering of targets from wideband SAR measurements. We first develop a scattering model for the response of an ideal dihedral when interrogated by a wideband radar. We formulate a stochastic detection problem based on this model and Gaussian clutter models. We investigate the performance of three detectors, the conventional imaging detector, a generalized likelihood ratio test (GLRT) detector based on the dihedral anisotropic scattering model, and a sum-of- squares detector motivated as a computationally attractive alternative to the GLRT test. We also investigate the performance degradation of the GLRT detector when using truncated angle response filters, and analyze detector sensitivity to changes in target length. Finally, we present initial results of angular matched filter detection applied to UWB radar measurements collected by the Army Research Laboratory at Aberdeen Proving Grounds.

We develop a theoretical bound on the receiver operating characteristics (ROC) associated with a target detector that bases its decision on the spatial distribution of extracted peak locations in synthetic-aperture (SAR) imagery. There are three basic steps to our analysis. In the first step, we formulate statistical models for both target and clutter regions of interest that have passed through a pre-screening stage. From these models, we then infer a corresponding detection procedure, which takes the form of a generalized likelihood ration test (GLRT). Finally, in the third step, we again use our statistical models to determine the ROC performance of the GLRT. This third step is where we believe our primary technical innovation lies. In the presence of uncertainty regarding target type and pose, it is generally difficult to obtain analytical performance expressions. We circumvent this problem by treating statistically the database of target exemplars. This approach has tow benefits. First, it suppresses the details of the database of exemplars, alleviating the need to specify the spatial configuration of points for every database entry. second, it leads to analytical performance expressions, which can be calculated with considerably less effort than is possible with Monte Carlo simulation.

The Gaussian form of the Bhattacharyya distance measure is being used by some in the automatic target recognition (ATR) community to select features and to estimate an upper performance bound for ATR algorithms. One reason for the popularity of this measure is that it is readily computed. This paper shows through both empirical and analytic results the inadequacy of this metric. Empirical results are obtained by processing ADTS field data through both the Gaussian form of the Bhattacharyya distance and a nonparametric error estimation scheme. Analytic results are obtained by deriving the Gaussian form of the Bhattacharyya distance metric for distributions other than Gaussian. These results show that the Gaussian form of the Bhattacharyya distance cannot be trusted to provide a reliable upper performance bound. Additional empirical and analytic results show by using a nonparametric performance estimator that when the data is transformed to be more Gaussian the Bhattacharyya metric gives better performance estimates. The transformations discussed are the power transform and a mode seeker that decomposes the data into Gaussian modes. A conclusion is that tools can be and should be developed that improve the utility of the Bhattacharyya metric mainly since they provide useful information about the distribution of the data. The major conclusion is that even with these tools nonparametric error estimation techniques are superior. The nonparametric performance bounds are more reliable, and the proper use of the Bhattacharyya metric depends upon considerable knowledge of the data distributions.

This paper discusses work toward a fundamental, algorithm- independent view of the ATR performance that can be achieved using SAR data. Such ATR performance predictions are intended to enable evaluation of performance tradeoffs for SAR designs, including both parameter selections and added domains of SAR observation, such as 3D, full polarimetry, multiaspect and/or multifrequency. In the paper we evaluate the classification error for two tactical targets using a Monte Carlo technique. A number of approximations are made and are detailed in the paper. Data on target signatures come from pencil-beam laser data and target photographs, which determine shadowing of the ground clutter. A single aspect angle is used for each target in the initial results. A layer of radar netting is modeled on both targets. This information is used as 'ground truth', to compute the average power that would be seen in each pixel of a SAR image, for each target. SAR image trials are then generated using independent Rayleigh amplitude fades in each pixel. In an optimal Bayesian fashion, the smaller of the probabilities of target (T1) or T2 given the trial image data is the error probability for that trial. An average over the Monte Carlo image trials yields the overall classification error probability. Comments are given on reducing the number of required trials in such a Monte Carlo. THree results of the work are shown. First, a tradeoff is made of ATR performance versus SAR resolution. ATR improves as the resolution is made finer, and physical reasons for this are discussed. Second, the relative ATR utility is determined for those pixels where at least one target has scatterer as compared with those pixels where the targets differ only in the degree to which they shadow the ground clutter. Third, an early analytical result is given for interferometric SAR, showing the physical reason behind the potential of height-sensing SAR to improve ATR - the possibility of canceling the background response becomes an important factor. Finally, the ability to make absolute performance predictions versus relative predictions is discussed, with the conclusion that relative predictions are more feasible at this time.

Many automatic target recognition systems work by matching test profiles against profile templates from various parts of the targets. These templates are generally constructed for uniform-width sub-targets constructed out of the targets being considered. In this paper we present a heuristic search based algorithm for constructing optimal-sized sub- targets which may be of varying widths. We show the improvements that ar obtained by constructing templates for sub-targets identified by the search algorithm. The reduction in mean square errors is reported and also the performance of two classification test runs.

Operational Synthetic Aperture Radar (SAR) based automatic target recognition (ATR) systems will encounter a wide range of target types, some of which will be variants of pre- mission training targets. Previous work measured classification performance when training and testing on different vehicle variants and assessed intra-class separability based on an empirical estimate of the mean square error (MSE) probability density function. This research showed a significant degree of intra-class signature variability for selective targets, resulting in a difficult ATR problem. The benefits of using mixture templates were demonstrated with respect to classification performance as well as pose prediction. This paper extends this analysis by considering the signature variability attributed to extended operating conditions such as depression angle and articulation. Furthermore, it demonstrates improved performance robustness is possible using an MSE classifier with appropriate normalization and segmentation. Additionally, a simple technique for minimizing the impact of localized error sources on MSE algorithms is discussed. Finally, error surfaces associated with missed classifications are shown to b similar in both space and amplitude, suggesting finer target discrimination may require improved feature sets and or adaptive refinement algorithms for handling both deterministic and random error sources associated with the observation to template.

The performance of a model-based automatic target recognition (ATR) engine with articulated and occluded objects in SAR imagery is characterized based on invariant properties of the objects. Using SAR scattering center locations as features, the invariance with articulation is shown as a function of object azimuth. The basic elements of our model-based recognition engine are described and performance results are given for various design parameters. The articulation invariant properties of the objects are used to characterize recognition engine performance, in terms of probability of correct identification as a function of percent invariance with articulation. Similar results are presented for object occlusion in the presence of noise, with percent unoccluded as the invariant measure. Finally, performance is characterized for occluded articulated objects as a function of number of features that are used. Results are presented using 4320 chips generated by XPATCH for 5 targets.

An attributed scattering center model exploits scattering phenomenology that is not accessed through traditional SAR image formation. Frequency, aspect, and polarization dependent scattering behaviors are jointly processed to provide a concise, descriptive, high resolution analysis of regions of interest. Used in conjunction with other features such as shadows, context, and image texture, attributed scattering center features hold promise for both feature- based and model-based automatic target recognition systems. In this conference paper, we present a parametric model for radar scattering as a function of frequency and aspect angle; the model is suggested by high-frequency monostatic far-field scattering solutions provided by the geometrical theory of diffraction and physical optics. The scattering model is used for analysis of synthetic aperture radar data. The estimated parameters provide a concise, physically relevant description of measured scattering for use in target recognition, data compression and scattering studies.The scattering model may be fit to either complex- valued imagery or to radar phase history data using a nonlinear least-squares estimator. Statistical analysis of the scattering model serves to characterize uncertainty to estimated scattering parameters. Feature estimation performance bounds are evaluated for X-band, K-band, and ultra wideband synthetic aperture radar scenarios.

This paper discusses enhancements of a recently developed system for robust, obscured object recognition by means of partial evidence reconstruction from object restricted measures (PERFORM). PERFORM employs a partial evidence accrual approach to form both an object identity metric and an object pose estimate. Recent enhancements of PERFORM resulted in significant performance improvements over those reported in our previous publications. Partial evidence information is obtained by applying several instances of the authors' linear signal decomposition/direction of arrival (LSD/DOA) pose estimation technique. The LSD/DOA ATR system avoids search techniques and is capable of detection and classification of possibly articulated, multiple objects with many degrees of freedom. Development of PERFORM was motivated by the fact that pose estimation in the LSD/DOA method is primarily degraded in practice by the presence of background clutter in the pose estimation filter's region of support. The use of several independent pose estimators based upon LSD/DOA's reciprocal basis set filters constructed for overlapping sub-regions of the object, allow for clutter independent pose estimation, and robust detection of obscured targets. Recent enhancements of the partial evidence fusion have been focused on introducing subhypothesis decisions related to the occupancy of the various subregions by target or clutter and fusion of information for target occupied regions. Presented results include receiver operating characteristic curves for SAR targets embedded in clutter with and without partial obscuration.

The Wright Lab Combat Information Division's system-oriented HRR automatic recognition program plans to develop and mature advanced air-to-ground high range resolution (HRR) automatic target recognition (ATR) capabilities for transition into suitable operational Air Force airborne platforms. ATR technologies have historically been template and feature based and generally provide good performance but at the expense of robustness. However, because of the need for robust ATR performance in an operational scenario, model-based ATR is being considered for possible future transition. This paper presents preliminary results on research dealing with what is commonly referred to as the 'training on synthetic - testing on measured' problem (TOS- TOM) in which algorithm trained on synthetic signatures often do not perform well when tested on measured data. The data trajectory realignment approach proposed in this paper makes the claim that poor TOS-TOM performance is the results of sufficient differences between the trajectories of synthetic data nd their measured counterpart. An adaptive model was developed which captures the trajectory relationship between synthetic and measured data and adaptively estimates the model parameters needed for realignment. Surrogate data sets were used for the preliminary proof of concept experiments and initial 2 class results showed more than 20 percent increase in probability of correct classification.

Squint side-looking strip imaging mode SAR has a variable squint angle from the side-looking direction, and has the capabilities of 'advance imaging' and 'turn-back imaging' to the terrain. The conventional side-looking mode SAR is only a special example of the squint mode SAR. Because of the large range migration and serious range-azimuth coupling terms, the imaging processing of squint mode SAR is a intrinsically 2D phenomenon. In this paper, different algorithms, which can be used for the imaging processing of squint mode SAR, are compared with each other in terms of their focusing quality and their ability to handle the large range migration and the space-variance of the correlation kernel of the squint side-looking mode SAR. And their ability of real-time imaging are also discussed. The algorithms contained here are 2D FFT method, fast polynomial transform method, and the direct correcting method based on range-Doppler focusing algorithms. Other new methods are also discussed here briefly.

This paper focuses on the estimation of the parameters of moving target of SAR. Basin on the analysis of the algorithm of cross-correlation of azimuth power spectra which is proposed by J.R. Moreira, this paper puts forwards a new kind of algorithm in time domain. Compared with cross- correlation algorithm, this new algorithm has the advantage of less computation loads. By computer simulation, this paper compares and analyzes these two algorithms. The estimation results in different SNR and different velocities are given. All these validate the effectiveness of this new algorithm.

Synthetic aperture radar (SAR) is getting more and more wide use in many areas. To lower the bit-flow will benefit both the communication and storage of raw data. Megellan SAR used a block adaptive quantization (BAQ) method. It estimates the statistics of the source and attempts to match the quantizer to the observed time-varying statistics. Because of its simplicity and effectiveness, many other SAR also used the BAQ algorithm. In this paper we use the polar quantizer instead of the quadrature quantizer in BAQ. Two kinds of polar quantizer are introduced and tested with X-SAR raw data.

An efficient tomographic algorithm is studied according to the high resolution imaging capability on small target terrain during the spotlight mode SAR procedure in this paper. By simplifying the procedure of Fourier reconstruction, reducing the A/D rat and PRF, implementing initial phase compensation, facilitating the hardware materialization, this tomographic algorithm for spotlight mode SAR greatly benefits the imaging processing efficiency. Necessary explanations, deductions, figures, tables and simulation results are given out.

We consider a land cover mapping method for polarimetric SAR data analysis. The method is based on a neural network whose input data are elements formed by the Stokes matrix. In this case, we must select a suitable combination of complex elements as a feature vector. After forming the probability density for each element and comparing the characteristics between JM distances, we determine a specific feature vector as the input for the network. As a result of experiments using SIR-C data, average accuracy for classification results is 86.40 percent, where (i) the 8D feature vector with backscattering coefficients and pseudo-phase differences between HH and VV from L and C bands and (ii) the competitive neural network with 8 input and 40 output neurons are simultaneously employed. In comparison, the proposed method outperforms other methods in average accuracy.

In this paper, a maximum likelihood (ML) classification algorithm is proposed to classify multi-look polarimetric SAR images. This algorithm considers a generalized multiplicative speckle model in which three texture factors are assumed to separately affect three polarization channels. We derive the ML estimation of the texture parameters for each polarization channel with the complex Wishart distribution of the multi-look speckle covariance matrix, and design the corresponding ML classifier according to the Bayesian criterion.BOth the texture class statistics and the discriminant function are given in simple closed forms. Further, a method for adaptively producing the a priori probabilities is also presented in order to improve the classification accuracy. This method utilizes the contextual information in a forward procedure, and does not need any iteration. With the NASA/JPL L-band 4-look polarimetric SAR data, the effectiveness of the presented classification algorithm is demonstrated, and using of the adaptive a priori probabilities is shown to result in improved classifications.

In the processing of interferometric SAR data, a number of experiences have been reported in the literature and various methods for the processing are available. Among the factors crucial to the accuracy of altitude mapping is the phase unwrapping process, which recovers the absolute phase value from the wrapped phase image. In this paper, sensitivity studies of the unwrapped phase error and the height inversion error are described using the simulated interferometric SAR images first. Then, the phase unwrapping algorithms are applied to real SAR data (JERS-1 SAR Mt Unzen data), and it is found that the accurate height is obtained by the branch-cut method even though the large phase noise of JERS-1 SAR data due to a long revisit time. Next, the data of Mt. Unzen obtained by interferometry using the SAR data taken after the eruption is compared with the DEM data of the same scene taken before the eruption. The result shows the consistency with the height changes which was found by geological survey after the eruption, and it is found that the JERS-1 SAR data can extract the topographic changes by differential SAR interferometry.

The main problems of detecting and imaging of moving targets with airborne synthetic aperture radar (SAR) are addressed. Novel methods are developed to estimate the two important parameters in imaging of moving targets -- the linear FM rate and Doppler frequency center of received echoes. There are extended Wigner-Ville distribution and Doppler frequency center estimation based on the WIgner-Ville distribution. So these parameters can be used to form synthetic aperture with respect to each moving object and produce high resolution image. Some estimated results and diagram of process are given.