Radar high-resolution range profile (HRRP) is very sensitive to time-shift and target-aspect variation; therefore, HRRP-based radar automatic target recognition (RATR) requires efficient time-shift invariant features and robust feature templates. Although higher order spectra are a set of well-known time-shift invariant features, direct use of them (except for power spectrum) is impractical due to their complexity. A method for calculating the Euclidean distance in higher order spectra feature space is proposed in this paper, which avoids calculating the higher order spectra, effectively reducing the computation complexity and storage requirement. Moreover, according to the widely used scattering center model, theoretical analysis and experimental results in this paper show that the feature vector extracted from the average profile in a small target-aspect sector has better generalization performance than the average feature vector in the same sector when both of them are used as feature templates in HRRP-based RATR. The proposed Euclidean distance calculation method and average profile-based template database are applied to two classification algorithms [the template matching method (TMM) and the radial basis function network (RBFN)] to evaluate the recognition performances of higher order spectra features. Experimental results for measured data show that the power spectrum has the best recognition performance among higher order spectra.

Radar HRRP target recognition based on higher order spectra

Computer vision systems today fail frequently. They also fail abruptly without warning or explanation. Alleviating the former has been the primary focus of the community. In this work, we hope to draw the community's attention to the latter, which is arguably equally problematic for real applications. We promote two metrics to evaluate failure prediction. We show that a surprisingly straightforward and general approach, that we call ALERT, can predict the likely accuracy (or failure) of a variety of computer vision systems--semantic segmentation, vanishing point and camera parameter estimation, and image memorability prediction--on individual input images. We also explore attribute prediction, where classifiers are typically meant to generalize to new unseen categories. We show that ALERT can be useful in predicting failures of this transfer. Finally, we leverage ALERT to improve the performance of a downstream application of attribute prediction: zero-shot learning. We show that ALERT can outperform several strong baselines for zero-shot learning on four datasets.

/pdf/predicting-failures-of-vision-systems-2jquibblsl.pdf

Predicting Failures of Vision Systems

In the statistical target recognition based on radar high-resolution range profile (HRRP), two challenging tasks are how to deal with the target-aspect, time-shift, and amplitude-scale sensitivity of HRRP and how to accurately describe HRRPs statistical characteristics. In this paper, based on the scattering center model, range cells are classified, in accordance with the number of predominant scatterers in each cell, into three statistical types. After resolving the three sensitivity problems, this paper develops a statistical model comprising two distribution forms, i.e., Gamma distribution and Gaussian mixture distribution, to model echoes of different types of range cells as the corresponding distribution forms. Determination of the type of a range cell is achieved by using the rival penalized competitive learning (RPCL) algorithm, while estimation for the parameters of Gamma distribution and Gaussian mixture distribution by the maximum likelihood (ML) method and the expectation-maximization (EM) algorithm, respectively. Experimental results for measured data show that the proposed statistical model not only has better recognition performance but also is more robust to noises than the two existing statistical models, i.e., Gaussian model and Gamma model.

A two-distribution compounded statistical model for Radar HRRP target recognition

Statistical modeling for radar high-resolution range profile (HRRP) is a challenging task in radar HRRP statistical recognition. Theoretical analysis and experimental results show that elements in an HRRP sample are statistically correlated and non-Gaussian distributed. First, this paper introduces three joint-Gaussian models, i.e., subspace approximation model, probability principal components analysis (PPCA) model and factor analysis (FA) model, into radar HRRP statistical recognition. Due to the experimental results, we can have the conclusion that the jointly non-Gaussian distributed HRRP samples approximately follow the joint-Gaussian distribution described by FA model. Therefore, we can apply FA model to radar HRRP statistical recognition rather than a joint-Gaussian mixture model, e.g., PPCA mixture model or FA mixture model, which is a more accurate choice for modeling non-Gaussian distributed correlations in multidimensional data but with high learning complexity and large computation burden, and the difficulty in the statistical modeling for HRRP samples is largely reduced. Second, this paper concerns model selection of FA model in radar HRRP statistical recognition, in which there are two issues, i.e., the partition of target-aspect frames and the determination of the number of factors in each frame. Based on the Akaike information criterion (AIC) and the Bayes' information criterion (BIC), an iterated algorithm for model selection is proposed in this paper, which can automatically give the optimal aspect-frame boundaries and determine the optimal number of factors in each aspect-frame. The recognition experiments based on measured data show that the proposed adaptive partition approach can further improve the recognition performance with higher recognition efficiency.

Radar HRRP Statistical Recognition: Parametric Model and Model Selection

Wideband radar can obtain much target information due to its high-range resolution. Because its size is smaller than the resolution of conventional radar, for an airplane, it can be regarded as a point. However, for wideband radar with a bandwidth of several hundred megahertz, the range resolution is less than a meter, and airplane echoes of one impulse form a high-resolution range profile (HRRP), which includes the shape information of a target that can be used for automatic target recognition (ATR). A range profile can be divided into many range resolution cells, each of which still contains the echoes of many scatterers. The complex amplitude of one range cell echo can be regarded as the sum of these echoes. A small aspect of change will lead to a big change of the complex amplitude, since, although the variation of the range of each scatterer to radar is very small, it is still large enough compared to the small wavelength and may cause a big phase change. Hence, the range profiles are very sensitive to the aspect angle, which makes the ATR based on HRRPs a well-known challenging problem. The properties of range profiles are investigated and a preprocessing scheme is presented to obtain stable range profiles for ATR.

Properties of high-resolution range profiles

Synthetic aperture radar (SAR) imaging and automatic target recognition (ATR) of moving targets pose a significant challenge due to the inherent difficulty of focusing moving targets. As a result, ATR of moving targets has received increased interest. High range resolution (HRR) radar mode offers an approach for recognizing moving targets by forming focused HRR profiles with significantly enhanced target-to-(clutter+noise) (T/(C+N)) via Doppler filtering and/or clutter cancellation. A goal of HRR ATR transition is the implementation and evaluation of algorithms exhibiting robustness under extended operating conditions (EOC). A subset of the moving and stationary target acquisition and recognition (MSTAR) data set was used to study 1D template-based ATR development and performance. Due to the unavailability of a statistically significant moving ground target data set, this approach was taken as an interim step in assessing the separability of ground targets when using range only discriminants. This report summarizes the data and algorithm methodology, performance results, and recommendations.

Automatic target recognition of time critical moving targets using 1D high range resolution (HRR) radar

This paper presents the performance of a multi-class, template-based, system-oriented High Range Resolution (HRR) Automatic Target Recognition (ATR) algorithm for ground moving targets. The HRR classifier assumes a target aspect estimate derived from the exploitation of moving target indication (MTI) mode target tracking to reduce the template search space. The impact of the MTI tracker target aspect estimate accuracy on the performance and robustness of the HRR ATR is investigated. Next, both individual and hybrid MTI/HRR and Synthetic Aperture Radar (SAR) model-based ATR algorithm results are presented. The hybrid ATR under consideration assumes the coordination of a multimode sensor to provide classification or continuous tracking of targets in a move- stop-move scenario. That is, a high-value moving target is tracked using the GMTI mode and its heading estimated. As the indicated target stops, the last GMTI tracker update is used to aid the SAR mode ATR target acquisition and classification. As the target begins to move again, the MTI-assisted HRR ATR target identification estimates are fused with the previous SAR ATR classification. The hybrid MTI/SAR/HRR ATR decision- level fusion provides a method for robust classification and/or continuous tracking of targets in move-stop-move cycles. Lastly, the baseline HRR ATR performance is compared to a QuickSAR (short dwell or non-square pixel SAR) ATR algorithm for varying cross-range resolutions.

Analysis of a 1D HRR moving target ATR

The statistical feature-based (StaF) classifier is presented for robust high range resolution (HRR) radar moving ground target identification. The target features used for classification are the amplitude and location of HRR signature peaks. The peak features are not predetermined using the training data but are extracted on-the-fly from the observed HRR profile and are different for each target observation. A classifier decision is made after statistical evidence is accrued from each feature and across multiple looks. Decision uncertainty is estimated using a belief-based confidence measure. Classifier decisions are rejected if the decision uncertainty is too high since it is likely that the observed HRR profile is not in the classifier's target database. Robustness is achieved by using only peak features rather than the entire HRR profile (much of which is low-level scatterers buried in noise or simply noise) and by rejecting decisions with high uncertainty.

Thomas Fister

Papers

Automatic target recognition of time critical moving targets using 1D high range resolution (HRR) radar

Analysis of a 1D HRR moving target ATR

Robust feature-based Bayesian ground target recognition using decision confidence for unknown target rejection