基于局部特性实现单帧图像小目标检测的研究

以目标成像点的扩散理论为基础,建立小目标在空域上的灰度特性模型,分析目标、背景和噪声的基本特性。由形态学开闭运算得到各像素位置的灰度变化值,再根据此值确定潜在目标区域。研究了各潜在目标区域的多方向多级梯度特征,实现了单帧图像的小目标检测。研究结果表明,该方法能够有效抑制不均匀背景杂波,增强目标信号,提高单帧亮暗点目标的检测能力。对于信噪比为0.89的图像,可获得34.74的信噪比增益。     

Multiresolution detection of coherent radar targets

We develop and investigate several novel multiresolution algorithms for detecting coherent radar targets embedded in clutter. These multiresolution detectors exploit the fact that prominent target scatterers interfere in a characteristic manner as resolution is changed, while multiresolution clutter signatures are random. We show, both on simulated and collected synthetic aperture radar data, that these multiresolution algorithms yield significant detection improvements over single-pixel, single-resolution constant false alarm rate (CFAR) methods that use only the finest available resolution.     

基于贝叶斯的高斯杂波背景下MIMO雷达自适应检测算法

对于集中式多输入多输出(Multiple-Input Multiple-Output, MIMO)雷达,该文研究了高斯杂波背景下的目标检测问题。该文假设杂波的协方差矩阵是未知随机的,且服从逆复Wishart分布,基于贝叶斯方法和广义似然比检验准则设计了两种新型自适应检测器。该文提出的贝叶斯检测器具有两个显著的优点:(1)不需要训练数据;(2)杂波的先验知识体现在设计方案中,从而提高了检测性能。仿真结果显示该文提出的贝叶斯检测器的检测性能优于目前常用的非贝叶斯检测器,特别是在发射波形采样数较少时。另外,该贝叶斯检测器在参数失配条件下的性能会有一定程度下降。      

基于流形等距映射的矩阵信息几何检测方法

雷达目标检测常面临复杂的杂波特性,经典的检测方法通常适合于某些特定的场景,当检测背景发生变化时,其检测性能急剧下降。为有效提升不同杂波背景下的检测性能,提出一种基于流形等距映射(ISOMAP)的矩阵信息几何检测器。该方法首先将信号检测问题转化为矩阵流形上两点之间的区分性问题;然后基于样本数据和流形等距映射原理,自适应地学习出矩阵流形的投影变换矩阵,将矩阵流形变换为可区分的低维流形,最大程度地保持每一个矩阵与其邻域内矩阵之间几何距离大小,增强矩阵流形的可分性;最后利用仿真杂波和实测数据对算法进行验证。实验结果表明,相比于经典的检测方法,所提方法能有效提升目标检测性能。     

基于检测前识别的MTS-CFAR检测方法

在复杂的强地杂波中检测静止目标时极易产生漏检,因此一直是检测领域中的难题。在双参数恒虚警(DP-CFAR)检测方法的基础上提出了基于检测前识别的多时间尺度恒虚警(MTS-CFAR)检测方法。在检测前基于各类型目标在多时间尺度统计结果上的不同,区分和识别运动目标、静止目标和杂波,减少杂波统计量和目标统计量的干扰,提高对静止目标的检测性能。以交通雷达应用为背景,通过仿真和实测数据定量分析和对比了MTS-CFAR、DP-CFAR、截断统计量恒虚警、改进型单元平均恒虚警等检测算法的性能,证明了本文算法对于静止目标检测效果的优越性和工程应用价值。     

基于数学形态学的红外图像小目标检测

针对红外成像跟踪系统的低信噪比、背景和噪声干扰严重的小目标图像,以及后续的目标识别处理需要目标的灰度信息的特点,设计了一种基于数学形态学的红外灰度图像小目标检测算法,并用FPGA硬件实现。实验表明,数学形态学滤波能够大大提高目标的信噪比,去除背景和噪声干扰,保留目标的灰度信息,满足系统实时性要求,是一种有效的红外灰度图像小目标检测算法。     

全卷积神经网络应用于SAR目标检测

在合成孔径雷达（SAR）图像目标检测中,由于场景杂波的复杂多变,对背景杂波统计模型估计难度增加,从而导致多数检测器容易受到背景杂波的干扰。针对如何避免场景杂波对目标检测干扰的问题,提出了一种基于全卷积神经网络的SAR目标检测模型。该模型将目标检测任务转化为像素分类问题,利用卷积神经网络对数据集中目标像素特征和背景杂波像素的先验信息进行自主学习,有效减少了虚警目标的数量;通过对目标及其阴影区域的联合检测,提高了目标的检测概率。对多个不同场景图像进行测试,实验结果表明提出的检测模型具有良好的检测性能和鲁棒性能,与传统恒虚警检测算法相比,在无需考虑背景杂波统计模型前提下有效降低了虚警概率。     

Finite-SNR Diversity–Multiplexing Tradeoff for Correlated Rayleigh and Rician MIMO Channels

A nonasymptotic framework is presented to analyze the diversity-multiplexing tradeoff of a multiple-input-multiple-output (MIMO) wireless system at finite signal-to-noise ratios (SNRs). The target data rate at each SNR is proportional to the capacity of an additive white Gaussian noise (AWGN) channel with an array gain. The proportionality constant, which can be interpreted as a finite-SNR spatial multiplexing gain, dictates the sensitivity of the rate adaptation policy to SNR. The diversity gain as a function of SNR for a fixed multiplexing gain is defined by the negative slope of the outage probability versus SNR curve on a log-log scale. The finite-SNR diversity gain provides an estimate of the additional power required to decrease the outage probability by a target amount. For general MIMO systems, lower bounds on the outage probabilities in correlated Rayleigh fading and Rician fading are used to estimate the diversity gain as a function of multiplexing gain and SNR. In addition, exact diversity gain expressions are determined for orthogonal space-time block codes (OSTBC). Spatial correlation significantly lowers the achievable diversity gain at finite SNR when compared to high-SNR asymptotic values. The presence of line-of-sight (LOS) components in Rician fading yields diversity gains higher than high-SNR asymptotic values at some SNRs and multiplexing gains while resulting in diversity gains near zero for multiplexing gains larger than unity. Furthermore, as the multiplexing gain approaches zero, the normalized limiting diversity gain, which can be interpreted in terms of the wideband slope and the high-SNR slope of spectral efficiency, exhibits slow convergence with SNR to the high-SNR asymptotic value. This finite-SNR framework for the diversity-multiplexing tradeoff is useful in MIMO system design for realistic SNRs and propagation environments     

LIKELIHOOD RATIO DETECTION OF SWERLING——Ⅱ TARGETS IN RICIAN CLUTTER

 The discrete time detection of Swerling Ⅱ targets in Rician clutter is investigated.The likelihood ratio (LR) detector is structured when the clutter steady part is coherent and noncoherent are developed respectively.The efficacy expressions of the detectors are derived.The calculation results show that the more the clutter statistics depart from the Gaussian situation,the better the detection performance of LR detectors is than that of square law detectors.The asymptotic relative efficiency (ARE) in case of clutter steady part being coherent can be twofold higher than that iu case of clutter steady part being noncoherent.     

The Rao and Wald Tests Designed for Distributed Targets with Polarization MIMO Radar in Compound-Gaussian Clutter

In high resolution radars, the distributed target is usually modeled as a few isolated points referred to multiple dominant scattering centers, while the clutter is a compound-Gaussian model. Additionally, the polarimetric diversity can be exploited to enhance detection performance. Motivated by extending the detection problem of multiple-input multiple-output (MIMO) radar to such cases, this paper mainly addresses distributed targets detection problem with polarization MIMO radar against a compound-Gaussian clutter dominated scenario with unknown covariance matrix. The adaptive detectors based on Rao and Wald criteria are studied, and a two-step design procedure is adopted. Specifically, the Rao and Wald tests are derived by assuming a known covariance matrix, and then a suitable estimation of the covariance matrix based on the secondary data is inserted into the derived detectors to make them fully adaptive. Some numerical results are presented together with a polarization generalized likelihood ratio test (GLRT), showing that the derived detectors provide excellent detection performance in spiky clutter for distributed targets, and that the polarimetric diversity can be exploited to improve detection performance. Overall, the Wald test performs the best.     

Efficient detection in hyperspectral imagery

Hyperspectral sensors collect hundreds of narrow and contiguously spaced spectral bands of data. Such sensors provide fully registered high resolution spatial and spectral images that are invaluable in discriminating between man-made objects and natural clutter backgrounds. The price paid for this high resolution data is extremely large data sets, several hundred of Mbytes for a single scene, that make storage and transmission difficult, thus requiring fast onboard processing techniques to reduce the data being transmitted. Attempts to apply traditional maximum likelihood detection techniques for in-flight processing of these massive amounts of hyperspectral data suffer from two limitations: first, they neglect the spatial correlation of the clutter by treating it as spatially white noise; second, their computational cost renders them prohibitive without significant data reduction like by grouping the spectral bands into clusters, with a consequent loss of spectral resolution. This paper presents a maximum likelihood detector that successfully confronts both problems: rather than ignoring the spatial and spectral correlations, our detector exploits them to its advantage; and it is computationally expedient, its complexity increasing only linearly with the number of spectral bands available. Our approach is based on a Gauss-Markov random field (GMRF) modeling of the clutter, which has the advantage of providing a direct parameterization of the inverse of the clutter covariance, the quantity of interest in the test statistic. We discuss in detail two alternative GMRF detectors: one based on a binary hypothesis approach, and the other on a "single" hypothesis formulation. We analyze extensively with real hyperspectral imagery data (HYDICE and SEBASS) the performance of the detectors, comparing them to a benchmark detector, the RX-algorithm. Our results show that the GMRF "single" hypothesis detector outperforms significantly in computational cost the RX-algorithm, while delivering noticeable detection performance improvement.     

动与静目标双重检测的手势检测方案及硬件设计

基于超声波的手势识别系统中引入手势检测方案一方面可以在无手势的时候不执行高复杂度的识别运算，有效地降低功耗；另一方面还可以分割出有效时间片段，提高手势识别的精度。为提升手势识别系统的性能，提出了一种手势检测方案并且以硬件实现了相应功能。采用动目标与静目标双重检测以及杂波图技术，提供手势识别系统在低功耗的检测模式以及高精度运算的识别模式间切换的条件，实现对环境静态杂波的跟踪和滤除，同时解决了传统单纯动目标检测在连续手势操作时存在的漏检问题。本方案由检测模式向识别模式切换的漏警率为0,识别模式向检测模式切换的虚警率为2%,模式切换的响应时间为733.47 ns。     

一种新的多频带相干雷达恒虚警检测算法

研究了复合高斯杂波环境中多频带相干雷达自适应恒虚警检测问题.利用Toeplitz矩阵的次对称性提出了一种新的多频带相干雷达恒虚警检测算法.所提出的算法对杂波的结构分量和杂波协方差矩阵都有恒虚警的性质,而且不需要杂波功率谱密度对称.该算法的检测性能优于以前提出的多频带归一化匹配滤波器算法.     

基于时频脊-Radon变换的海面小目标检测方法

针对海杂波背景下雷达对海面弱小目标探测技术难题,提出一种基于时频脊-Radon变换的帧平滑双特征检测方法。该方法首先采用分块白化方法对海杂波进行抑制,根据抑制后海杂波单元与目标单元在时频域二维平面呈现的不同能量分布特征,对时频脊做Radon变换并在变换域特征空间中提取出峰值和频宽特征。考虑到实际雷达采用的相干脉冲数通常较少（大多情况下为64个或更少）,易导致特征不稳定,进而影响海杂波与目标可分性,为此,通过多帧扫描历史数据和当前帧数据的综合应用,对特征做帧平滑处理以增强可分性。最后采用凸包分类算法,在双特征平面进行融合检测。经2级和4级海况实测数据验证,在同等参数条件下,本文检测方法相比已有基于时频三特征的检测方法、频域CFAR检测方法和分形特征检测方法具有更好的检测性能。      

Subspace-based adaptive generalized likelihood ratio detection

Subspace-based adaptive detection performance is examined for the generalized likelihood ratio detector based on Wilks' Λ statistic. The problem considered here is detecting the presence of one or more signals of known shape embedded in Gaussian distributed noise with unknown covariance structure. The data is mapped into a subspace prior to detection. The probability of false alarm is independent of the subspace transformation and depends only on subspace dimension. The probability of detection depends on the subspace transformation through a nonadaptive signal-to-noise ratio (SNR) parameter. Subspace processing results in an SNR loss that tends to decrease performance and a gain in statistical stability that tends to increase performance. It is shown that the statistical stability effect dominates the SNR loss for short data records, and subspace detectors can require substantially less SNR than full space detectors for equivalent performance. A method for designing the subspace transformation to minimize the SNR loss is proposed and illustrated through simulations     

红外图像中小目标检测研究

在获得的红外序列图像中,检测和跟踪微弱小目标一直是研究的重点。本文对小目标检测系统的背景抑制和目标检测算法和方案进行了调研。采用背景抑制算法以获得更高的信噪比,利用NP准则对图像进行分割,然后用识别算法对目标进行跟踪和识别。计算结果显示,抑制杂波后,采用序贯图像检测能够很好地增加探测概率。     

红外背景抑制与小目标检测算法

在有云层背景的红外序列图像中,检测和跟踪微弱小目标一直是研究的重点。介绍了一种简单可行的在云层背景下检测小目标的方案。算法中,首先进行背景抑制以获得更高的信噪比,接着是用门限分割以得到二值图像,最后依照目标在时间和空间运动的连续性,用一种管线法进行目标识别。实验结果表明算法十分有效。     

基于双邻域对比度的红外小目标检测算法

为了解决密集多目标检测中易造成的漏检问题,提出一种基于双邻域对比度的红外小目标检测算法。首先利用峰值搜索算法筛选出候选目标;再通过单尺度3层双邻域窗口遍历候选目标; 最后利用双邻域对比度模型计算候选目标区域的最小灰度对比度,并用对角梯度因子增强对比度和抑制杂波。结果表明,与5种对比方法相比,该方法的背景抑制因子和对比度增益分别平均提高4.7倍和1.8倍,有效地抑制了杂波,增强了目标。该研究能够准确地检测到相互接近的多个目标,对提高复杂背景下的多目标检测精度是有帮助的。     

A hybrid STAP approach to target detection for heterogeneous scenarios in radar seekers

Adaptive detection of moving targets on the sea is important for radar seekers. Recently, more attention has been paid to the deleterious effect of clutter heterogeneity on space-time adaptive processing (STAP) for pulse Doppler radar. Since secondary samples are no longer statistically independent and identically distributed (IID) in heterogeneous environments, this is subjected to a great challenge to target detection for radar seekers. Due to the fact that chaff jamming severely affects the performance degradation of target detection, the hybrid detection algorithm is proposed to suppress the sea clutter and chaff jamming. Firstly, the range cells can be classified into two regions according to the power, namely clutter region and hybrid region. Then we propose different algorithms to process two regions. The fixed point (FP) estimator is used to estimate the clutter covariance matrix in clutter region. While the power selected training (PST) algorithm is used to select the homogeneous secondary samples, and an algorithm based on two-step subspace projection for hybrid interference suppression is presented in hybrid region. Finally, the proposed Pareto-based generalized likelihood ratio test (PBGLRT) detector can detect the slowly moving targets in heterogeneous interference. Simulation results show that the PBGLRT detector outperforms both the low rank normalized adaptive match filter (LRNAMF) and normalized adaptive match filter (NAMF) detectors against interference heterogeneity.     

非高斯海杂波背景最优相干检测性能分析

非高斯海杂波背景下雷达目标检测一直是研究的重点。海杂波分布参数形式的多样化使得检测器对杂波平均功率的恒虚警特性无法保证,因此文中对描述非高斯海杂波的K分布、广义Pareto分布和逆高斯纹理复合高斯分布的参数形式进行了规范化,得到了规范化后的三种杂波分布对应的最优相干检测器。理论分析和实验验证了三个自适应检测器分别对杂波平均功率、目标多普勒频率和散斑协方差矩阵具有恒虚警特性。仿真实验表明:散斑协方差矩阵估计方法和辅助数据的数量均影响着自适应检测器的检测概率和虚警概率。