三种两样本非参量CFAR检测算法分析

介绍了3种两样本非参量CFAR检测算法的基本工作原理,利用实测未知统计概率分布海杂波数据对它们的检测性能进行了研究,并与参量CA-CFAR检测器进行了对比.研究表明:在强海杂波条件下,GS-CFAR检测器的检测性能最优;在弱海杂波条件下,Savage-CFAR检测器的检测性能最优;相比于CA-CFAR检测器,3种两样本...     

Polarization diversity detection in compound-Gaussian clutter

We present the problem of polarization diversity detection in compound-Gaussian clutter with unknown covariance matrix. To this end we assume that a set of secondary data, free of signal components and with the same covariance structure of the cell under test, is available. Due to the lack of a uniformly most powerful (UMP) detector we resort to a design procedure based upon the Rao and the Wald tests. Specifically we first derive the Rao and the Wald tests assuming that the covariance matrix is known, and then we substitute into the derived decision rules a suitable estimate of the clutter covariance. Interestingly, the newly proposed detectors share the constant false alarm rate (CFAR) property with respect to the texture statistical characterization. Moreover simulation results have shown that the Wald test based detector ensures a performance level higher than the Rao test. We have also conducted a further performance analysis, in the presence of real clutter data and in comparison with the previously proposed generalized likelihood ratio test (GLRT) based receivers, which highlights that, in general, the Wald test receiver outperforms its counterparts. Finally, since the newly proposed decision rules as well as the previously designed GLRTs do not ensure the CFAR property with respect to the clutter covariance matrix, we have developed a sensitivity analysis on the probability of false alarm (P/sub fa/), based on simulated clutter with covariance matrix estimated from real radar data. The results have shown that (P/sub fa/) is only slightly affected by variations in the clutter correlation properties and hence the CFARness is substantially achieved.     

Vector subspace detection in compound-Gaussian clutter. Part I: survey and new results

Deals with the problem of detecting subspace random signals against correlated non-Gaussian clutter exploiting different degrees of knowledge on target and clutter statistical characteristics. The clutter process is modeled by the compound-Gaussian distribution. In the first part of the paper, the optimum Neyman-Pearson (NP) detector, the generalized likelihood ratio test (GLRT), and a constant false-alarm rate (CFAR) detector are sequentially derived both for the Gaussian and the compound-Gaussian scenarios. Different interpretations of the various detectors are provided to highlight the relationships and the differences among them. In particular, we show how the GLRT detector may be recast into an estimator-correlator form and into another form, namely a generalized whitening-matched filter (GWMF), which is the GLRT detector against Gaussian disturbance, compared with a data-dependent threshold. In the second part of this paper, the proposed detectors are tested against both simulated data and measured high resolution sea clutter data to investigate the dependence of their performance on the various clutter and signal parameters.     

Data quantization effects in CFAR signal detection

In this paper, we investigate data quantization effects in constant false alarm rate (CFAR) signal detection. Exponential distribution for the input data and uniform quantization are assumed for the CFAR detector analysis. Such assumptions are valid in the case of radar for a Swerling I target in Gaussian clutter plus noise and a receiver with analog square-law detection followed by analog-to-digital (A/D) conversion. False alarm and detection probabilities of the cell averaging (CA) and order statistic (OS) CFAR detectors operating on quantized observations are analytically determined. In homogeneous backgrounds with 15 dB clutter power fluctuations, we show analytically that a 12-bit uniform quantizer is sufficient to achieve false alarm rate invariance. Detector performance characteristics in nonhomogeneous backgrounds, due to regions of clutter power transitions and multiple interfering targets, are also presented and detailed comparisons are given     

Sum-difference Reduced-dimensional Processing in SAR Image Domain and Statistical Analysis for Ground Moving Target Indication

In this article, a new reduced-dimensional adaptive processing algorithm based on joint pixels sum-difference data for clutter rejection is proposed. The sum-difference data are obtained by orthogonal projection of the joint pixels data of different synthetic aperture radar (SAR) images generated by a multi-satellite radar system. In the sense of statistical expectation, the sum-differ- ence data contain the common and different information of the SAR images. Therefore, the objective of clutter cancellation can be achieved by adaptive processing. Moreover, based on the residual image after clutter rejection, statistical analysis of constant false-alarm rate (CFAR) detection of moving targets is also presented. Simulation results demonstrate the effectiveness and robustness of the proposed algorithm even with heterogeneous clutter and image co-registration error.     

A CFAR Design for a Window Spanning Two Clutter Fields

When the heterogeneous clutter field spanning the spatial sampling sliding window can be modeled as two contiguous homogeneous clutter fields with the statistical parameters of each field unknown and independent from field to field and with the transition point between fields also not known, then the cell-averaging constant false alarm rate (CFAR) performance significantly degrades, yielding target masking effects and loss of false alarm regulation. For the same defined and encountered environment spanning the sliding window, the performance degradation effects are shown to be largely eliminated when a newly developed class of CFAR tests is employed. These tests are designated as heterogeneous clutter estimating CFARs (HCE-CFAR). The test initially involves the combined use of multiple hypothesis testing and maximum likelihood estimation procedures to estimate the statistical parameters of each of the two fields, and the transition point between them, and then makes use of the relevant estimated clutter field parameters to effect the final decision rule. HCE-CFAR designs are presented for both the cases when the contiguous fields have Rayleigh first-order probability distributions, and log-normal probability distribution. However, the focus of the development and the conducted performance evaluation is for the Rayleigh clutter cases.     

Analysis of some modified ordered statistic CFAR: OSGO and OSSOCFAR

It is necessary for automatic detection radars to be adaptive to variations in background clutter in order to maintain a constant false alarm rate (CFAR). A CFAR based on an ordered statistic technique (OS CFAR) has some advantages over the cell-averaging technique (CA CFAR), especially in clutter edges or multiple target environments; unfortunately the large processing time required by this technique limits its use. The authors present two new OS CFARs that require only ahlf the processing time. One is an ordered statistic greatest of CFAR (OSGO), while the other is an ordered statistic smallest of CFAR (OSSO). The OSGO CFAR has the advantages of the OS CFAR with only a negligible increment to the CFAR loss     

Low-Loss Almost Constant False-Alarm Rate Processors

Conventional normalizing constant false-alarm rate (CFAR) circuits use the same configurations when detecting targets in interference regions and in clear regions. The CFAR penalty incurred in the clear region can be reduced by using CFAR processors that recognize the region is clear so that normalization is not necessary. An analysis of the target-detection performance for a particular modified CFAR processor, for an active-radar sensor, and for a passive infrared (IR), or sonar, sensor is given. It is shown that the decreased CFAR penalty in the clear is coupled with an increase of false-alarm rate in the clutter regions.     

Vector subspace detection in compound-Gaussian clutter. Part II: performance analysis

For pt. I see ibid., vol. 38, no. 4, p. 1295 (2002). In this second part we deal with the problem of detecting subspace random signals against correlated non-Gaussian clutter modeled by the compound-Gaussian distribution. In the first part of the paper, we derived the optimum Neyman-Pearson (NP) detector, the generalized likelihood ratio test (GLRT), and a constant false-alarm rate (CFAR) detector; we also provided some interesting interpretations of them. In this second part, these detectors are tested against both simulated data and measured high resolution sea clutter data to investigate the dependence of their performance on the various clutter and signal parameters. Numerical examples concern a space-time adaptive processing (STAP) scenario and a ground-based surveillance radar system scenario.     

Matched subspace CFAR detection of hovering helicopters

A constant false alarm rate (CFAR) strategy for detecting a Gaussian distributed random signal against correlated non-Gaussian clutter is developed. The proposed algorithm is based on Scharf's matched subspace detector (MSD) and has the CFAR property with respect to the clutter amplitude probability density function (apdf), provided that the clutter distribution belongs to the compound-Gaussian family and the clutter covariance matrix is known to within a scale factor. Analytical expressions of false alarm and detection probabilities are derived. An application to the problem of detecting hovering helicopters against vegetated ground clutter is reported     

Analysis of CFAR processors in homogeneous background

Five different constant false alarm rate (CFAR) radar processing schemes are considered and their performances analyzed in homogeneous and nonhomogeneous backgrounds, the latter specifically being the multiple target environment and regions of clutter transitions. The average detection threshold for each of the CFAR schemes was computed to measure and compare the detection performance in homogeneous noise background. The exponential noise model was used for clear and clutter backgrounds to get closed-form expressions. The processor types compared are: the cell-averaging CFAR, the `greatest of' CFAR, the `smallest of' CFAR, the ordered-statistics CFAR, and a modified ordered-statistics processor called the trimmed-mean CFAR     

Further results on adaptive filtering with embedded CFAR

Among the few known adaptive filtering algorithms which have an embedded (integrated) constant false alarm rate (CFAR) performance feature, the generalized likelihood ratio (GLR) test algorithm has been found to be robust in non-Gaussian clutter. This paper examines the detection performance of the GLR algorithm in nonhomogeneous/nonstationary clutter environments which lead to nonidentical distribution of secondary (training) data. For two common types of nonhomogeneity, i.e., the so-called “signal contamination” and “clutter edge”, the asymptotic detection performance is derived and compared with simulations. These asymptotic results are relatively simple to use and they predict the GLR performance in nonhomogeneous environments quite well. The GLR performance loss due to the nonhomogeneity is also evaluated. It is found that the “generalized angle” between the desired and contaminating signal plays an important role in the study of the effects of signal contamination. It is also found that the performance degradation due to the clutter edge depends largely on the width of the clutter spectrum and target-clutter Doppler separation     

GLRT subspace detection for range and Doppler distributed targets

A generalized likelihood ratio test (GLRT) is derived for adaptive detection of range and Doppler-distributed targets. The clutter is modeled as a spherically invariant random process (SIRP) and its texture component is range dependent (heterogeneous clutter). We suppose here that the speckle component covariance matrix is known or estimated thanks to a secondary data set. Thus, unknown parameters to be estimated are local texture values, the complex amplitudes and Doppler frequencies of all scattering centers. To do so, we use superresolution methods. The proposed detector assumes a priori knowledge on the spatial distribution of the target and has the precious property of having a constant false alarm rate (CFAR) with the assumption of a known speckle covariance matrix or by the use of frequency agility.     

Radar CFAR Thresholding in Clutter and Multiple Target Situations

Radar detection procedures involve the comparison of the received signal amplitude to a threshold. In order to obtain a constant false-alarm rate (CFAR), an adaptive threshold must be applied reflecting the local clutter situation. The cell averaging approach, for example, is an adaptive procedure. A CFAR method is discussed using as the CFAR threshold one single value selected from the so-called ordered statistic (this method is fundamentally different from a rank statistic). This procedure has some advantages over cell averaging CFAR, especially in cases where more than one target is present within the reference window on which estimation of the local clutter situation is based, or where this reference window is crossing clutter edges.     

Ll-CFAR: a flexible and robust alternative

A new family of constant false alarm rate (CFAR) processors is introduced. An Ll-CFAR forms its noise power estimate by linearly filtering ranked samples from the reference set; the weights of this combination, however, depend not only on the rank, but also on the relative proximity of the sample to the cell under test. From the class of Ll-CFARs may be chosen members which effectively censor spurious targets; members which exhibit impressive control of false alarm in the presence of a clutter edge; and members which are robust against both such inhomogeneities. While the design of such schemes is involved, their implementation is not significantly more burdensome than that of plain ordered statistic CFAR (OS-CFAR). After a discussion of the stochastic training of Ll-CFAR, the performance is thoroughly assessed under the most commonly encountered instances of environmental conditions, and compared with those of classical CFAR techniques     

A Model for Generating Synthetic VHF SAR Forest Clutter Images

We propose a model for generating low-frequency synthetic aperture radar (SAR) clutter that relates model parameters to physical characteristics of the scene. The model includes both distributed scattering and large-amplitude discrete clutter responses. The model also incorporates the SAR imaging process, which introduces correlation among image pixels. The model may be used to generate synthetic clutter for a range of environmental operating conditions for use in target detection performance evaluation of the radar and automatic target detection/recognition algorithms. We derive a statistical representation of the proposed clutter model's pixel amplitudes and compare with measured data from the CARABAS-II SAR. Simulated clutter images capture the structure and amplitude responses seen in the measured data. A statistical analysis shows an order of magnitude improvement in model fit error compared with standard maximum-likelihood (ML) density fitting methods.     

Discrimination Between Log-Normal and Weibull Clutter

Two simple tests are presented for classifying a set of clutter samples into either the log-normal or Weibull distribution. The results obtained by Monte Carlo simulation have shown that both of these tests are only slightly inferior to the test based on the ratio of maximized likelihoods. An application to constant false-alarm rate (CFAR) processing is also discussed.     

非相干Rice杂波中的恒虚警检测

 地杂波的统计特性常常可以用Rice模型来描述,其物理基础是认为地杂波由一些大的固定散射体引起的稳定分量和大量小的随机分布的运动散射体引起的瑞利起伏分量所合成。文献[2]研究了稳定分量不相干时Rice杂波中离散时间最佳检测的估值器——相关器结构,但无显式解,实现有困难。文献[3]导出了Rice杂波中SwerlingⅡ目标的离散时间检测的似然比检测器结构。在此基础上,本文给出了一种修正平方律结构的似然比检测器,并和通常的平方律检测器作了性能比较。     

Performance of MAX family of order-statistic CFAR detectors

The MAX family of constant-false-alarm-rate (CFAR) detectors is introduced as a generalization of the greatest of CFAR (GO-CFAR) or MX mean-level detector (MX-MLD). Members of the MAX family use local estimators based on order statistics and generate both a near-range and a far-range noise-level estimate. Local estimates are always combined through a maximum operation; this insures false-alarm control at clutter edges. At the same time, order-statistic-based estimators result in a high-resolution detector. A complete detection analysis is provided for SWII targets and a reference channel contaminated by large outliers. Results are presented for the MX censored MLD (MX-CMLD) operating in clutter. The MX order statistic detector (MX-OSD) based on only a single-order statistic per window, is analyzed, and curves showing the required threshold, CFAR loss, optimum censoring point, and signal-to-noise ratio (SNR) loss in the presence of outliers are given. Simulations are used to compare the dynamic responses of various MX-OSD detectors in a clutter and a multiple-target environment     

Target prescreening based on a quadratic gamma discriminator

This work presents the development, analysis and validation of a new target discrimination module for synthetic aperture radar (SAR) imagery based on an extension of gamma functions to 2-D. Using the two parameter constant false-alarm rate (CFAR) stencil as a prototype, a new stencil based on 2-D gamma functions is used to estimate the intensity of the pixel under test and its surroundings. A quadratic discriminant function is created from these estimates, which is optimally adapted with least squares in a training set of representative clutter and target chips. This discriminator is called the quadratic gamma discriminator (QGD). The combination of the CFAR and the QGD was tested in realistic SAR environments and the results show a large improvement of the false alarm rate with respect to the two-parameter CFAR, both with high resolution (1 ft) fully polarimetric SAR and with one polarization, 1 m SAR data