-
摘要: 极化合成孔径雷达(PolSAR)使用二维脉冲压缩技术获取高分辨力极化信息图像,目前已广泛应用在军事侦察、地形测绘、环境与自然灾害监视、海上舰船目标检测等领域。如何解决复杂海杂波的建模与参数估计、慢小目标检测、密集目标检测等问题仍然是当前PolSAR图像舰船目标检测的难点。该文归纳梳理了PolSAR图像舰船目标检测的4类主流方法:极化特征目标检测方法、慢速运动目标检测方法、舰船目标尾迹检测方法以及基于深度学习的目标检测方法等,同时给出了各类方法所存在的问题以及可能的解决方法,并预测了其未来研究重点和发展趋势。Abstract: Polarimetric Synthetic Aperture Radar (PolSAR) uses two-dimensional pulse compression to obtain high-resolution images containing polarimetric information. PolSAR has been widely used in military reconnaissance, topographic mapping, environmental and natural disaster monitoring, marine ship detection, and related fields. Addressing the problems associated with sea-clutter modelling and parameter estimation, slow and small target detection, dense target detection, as well as other issues, still remains a challenge in PolSAR ship detection. In this paper, four main classes for PolSAR ship detection are summarized: target polarimetric feature detection, slow and small target detection, ship wake detection, and deep learning detection. In addition, the possible solutions to existing problems in each class are given, and their future development trends are predicted, which can provide some valuable suggestions for interested researchers.
-
Key words:
- Polarimetric SAR /
- Ship detection /
- Performance comparison /
- Development history /
- Future trend
-
1. 引言
机载雷达前视成像在态势感知、自主导航、地形跟踪和地形跟随等方面有着广泛的应用[1−3]。由于扫描雷达具有连续和广域观测的优势,因此常用于前视成像。然而与其他成像技术如双基前视合成孔径雷达成像方法相比,扫描雷达前视成像存在角度分辨率低的问题[4−7]。
近年来,研究学者提出了多种超分辨方法来获得高角度分辨率[8,9]。在超分辨方法中,扫描雷达回波信号可以看作天线方向图与目标散射系数之间的卷积。目标散射系数可以通过反卷积的方式得到,但是因为天线方向图的低通特性,直接进行反卷积是一个病态问题[10]。在矩阵分析中,天线方向图矩阵中小的奇异值求逆会提高噪声的幅值。根据这个特点,文献[11]提出了截断奇异值分解(Truncated Singular Value Decomposition, TSVD)方法,通过舍弃小的奇异值来抑制噪声。正则化方法也是一个很好地解决病态问题的方法,其核心是选择合适的正则化项[12−14]。在文献[15]中,作者提出了Tikhonov正则化,其引入L2范数项来抑制噪声。除此之外,谱估计方法也可以用来提高角度分辨率[16−18]。电子科技大学张永超等人[19−21]将迭代自适应方法(Iterative Adaptive Approach, IAA)用于前视超分辨成像,并证明了IAA方法可以抑制噪声并提高角度分辨率。与整个成像场景相比,感兴趣成像目标是稀疏的,因此稀疏成像被引入到机载雷达前视超分辨成像中[22−25]。在文献[26]中,作者提出了一个贝叶斯稀疏前视成像方法,该方法利用拉普拉斯分布来描述目标的稀疏特性。成像结果显示拉普拉斯分布可以很好地拟合目标,提高前视成像的角度分辨率。当飞机在船舶上自主着陆时,船舶的轮廓信息尤为重要。文献[27]将全变差(Total Variation, TV)和L1范数作为正则化约束项来实现前视成像。该方法在提高雷达前视成像角度分辨率的同时,很好地保留了目标的轮廓信息。
相比于提高前视成像的角度分辨率,如何降低前视成像的计算复杂度也是当前研究的一个热点[28−30]。文献[31]提出将雷达回波信号从时域转换到频域,利用特普利茨矩阵的低位移秩特点,采用Gohberg-Semencul分解实现了矩阵的快速求逆。因为矩阵的维度影响前视成像的实时性,文献[32]采用低秩近似思想对回波信号降维,并采用低秩迭代自适应方法(Low Rank Iterative Adaptive Approach, LRIAA)实现快速前视成像。文献[33]提出了一个二维前视成像处理方法,通过对整体的回波信号进行处理,避免了逐距离单元进行前视成像带来的计算负担,提高了前视成像的实时性。
在前视成像中,雷达回波可能会受到电磁脉冲干扰,或者受设备性能不稳定影响,导致雷达回波数据出现异常[34,35]。传统的成像方法假设噪声的统计分布服从高斯分布。在面对异常值时,异常值可能会被重构为虚假目标,影响前视成像的准确性。在之前的工作中,我们提出采用学生t分布来拟合雷达回波数据异常时的噪声[35]。相比于高斯混合模型,基于学生t分布的超分辨方法避免了估计过多的系统参数。但是在估计过程中基于学生t分布的方法依然面临高计算复杂度问题。
本文提出了一种机载雷达超分辨方法实现回波数据异常时的快速前视成像。首先,为了更好地拟合噪声和异常值,我们引入对异常值更加鲁棒的学生t分布。为了更好地重构目标,引入拉普拉斯分层先验来模拟目标的稀疏特征。然后基于期望最大化方法对成像参数进行估计。受TSVD方法的启发,将截断酉矩阵引入目标散射系数的估计公式中。通过矩阵变换降低了求逆矩阵的尺寸从而降低了参数估计的计算复杂度。仿真结果表明本文提出的快速方法可以通过更短的时间实现回波数据异常情况下的前视超分辨成像。
2. 机载雷达前视成像回波模型
图1为扫描雷达前视成像的几何示意图。图中机载平台距离地面高度为H,搭载着雷达以速度v向前飞行,天线以扫描速度ω逆时针扫描成像区域。初始时刻,载机位于点A所在位置。此时雷达与目标P1之间的初始距离记为R0,目标P1的方位角记为θ0,俯仰角记为φ0。经过t时刻,载机飞行至B,此时雷达与目标P1的方位角为θ(t),雷达与目标P1之间的瞬时距离记为R(t)。
假设雷达发射线性调频信号
s(τ)=rect(τ/T)exp(jπμτ2)exp(j2πfcτ) (1) 其中,τ为快时间,T为脉冲持续时间,μ为脉冲调制频率,fc为脉冲载波频率,rect(⋅)为窗函数。
对于目标P1,经过脉冲压缩和运动补偿后的回波信号可写为
s(τ,t)=σh(t−t0)sinc[Br(τ−2R0c)]⋅exp(−j4πλR(t)) (2) 其中,σ表示目标P1的散射系数。h(⋅)表示天线方向图调制函数。Br为信号带宽。λ为信号波长。式(2)中指数项exp(−j4πR(t)/λ)为多普勒频移项。瞬时距离R(t)可写为
R(t)=√R20+(vt)2−2R0vtcosθ0cosφ0≈R0−vtcosθ0cosφ0 (3) 在某一距离单元,回波信号可写为
y=(F⊙H)σ+n (4) 其中,y=[y1,y2,⋯,yM]T为该距离单元上的回波向量。M为回波信号的长度。σ=[σ1,σ2,⋯,σM]T为目标散射向量。n=[n1,n2,⋯,nM]T为成像噪声。F为多普勒相位矩阵。H为由天线方向图向量h=[h1,h2,⋯,hL]T组成的方向图矩阵。L为天线方向图长度。
F=[11⋯1ej2πfd(θ1)PRFej2πfd(θ2)PRF⋯ej2πfd(θM)PRF⋮⋮⋱⋮ej2πfd(θ1)M−1PRFej2πfd(θ2)M−1PRF⋯ej2πfd(θM)M−1PRF]M×M (5) H=[h⌈L/2⌉h⌈L/2⌉−1⋯h1h⌈L/2⌉+1h⌈L/2⌉⋱h1⋮⋮⋱⋱⋱hL⋮⋱⋱h1hL⋱⋱h2⋱⋱⋱⋮hLhL−1⋯h⌈L/2⌉]M×M (6) fd(θm)为多普勒中心频率。PRF为脉冲重复频率。⌈⋅⌉为向上取整操作。
fd(θm)=2vcosθmcosφλ (7) φ为此距离单元上的俯仰角。为了获取目标散射向量的幅值和位置信息,本文仅考虑式(4)中的幅值信息,即
|y|=|(F⊙H)σ|+|n| (8) 方便起见,将式(8)写为
s=Hx+n (9) 其中,s=[s1,s2,⋯,sM]T为回波幅值向量,x=[x1,x2,⋯,xM]T为目标散射幅值向量。
3. 回波数据异常时的快速前视超分辨方法
本节提出了快速超分辨方法实现回波数据异常时的前视成像。首先,采用学生t分布来拟合回波数据异常时的噪声。接着引入拉普拉斯分层分布模拟目标的稀疏分布。然后采用期望最大化方法实现对成像参数的估计。最后将截断的酉矩阵引入到目标散射系数的估计公式中,并通过矩阵变换对目标估计公式进行重写,降低估计过程中的计算复杂度。
3.1 贝叶斯模型
在前视成像过程中,雷达边发射波束边扫描前方场景,通过接收目标的扫描信号对前视场景进行成像。然而在成像场景中可能存在无意的窄带电磁干扰,如非合作脉冲雷达发射的信号。在这种情况下,回波信号的某些方位角上将会出现异常值。此外,雷达受设备性能的影响,如应答器性能、通信链路性能、信号时间戳抖动/量化等多种因素。在这种情况下,整个回波信号中会在不同的方位向距离向随机出现异常值。
受异常值影响,成像噪声变为系统固有高斯白噪声和异常值的组合。文献[36]指出,与高斯分布相比,学生t分布可以容忍回波信号中少数异常值的存在
p(n|μ,η,v)=Γ(M+ν2)Γ(ν2)(ηπν)M2⋅[1+(n−μ)Tη(n−μ)ν]−M+ν2 (10) 其中,μ为学生t分布的均值,η为学生t分布的精度。ν为自由度,当ν=1,学生t分布变为柯西分布,当ν→∞,学生t分布变为高斯分布。Γ(⋅)为Gamma函数。
图2给出了高斯分布和学生t分布的概率密度图比较。图中高斯分布的均值为0、方差为1.13,学生t分布的均值为0、精度为1、自由度为2。从图中可以看出,虽然高斯分布与学生t分布的峰值近似,但学生t分布拥有更高的拖尾,更高的拖尾意味着学生t分布可以容忍回波信号中少量的异常值存在,对异常值更加鲁棒。为了直观地展示学生t分布对异常值的鲁棒性,我们分别拟合了高斯数据和混合少量异常值的高斯数据。直方图拟合曲线如图3所示,其中图3(a)为高斯分布和学生t分布对高斯数据的直方图拟合曲线,图3(b)为高斯分布和学生t分布对混合少量异常值数据的直方图拟合曲线。相比于图3(a)中的拟合曲线,图3(b)中高斯分布的拟合曲线受异常值的影响变化较大,学生t分布的拟合曲线变化较小,说明学生t分布比高斯分布对异常值的鲁棒性更好。
同时,学生t分布也可由高斯分布和Gamma分布得到
p(n|μ,η,v)=∫∞0N(n|μ,(ηΣ)−1)G(Σ|ν2,ν2)dΣ (11) 其中,η和Σ=diag(ε1,ε2,⋯,εM)共同表示高斯分布的逆协方差矩阵。N(⋅)为高斯分布,G(⋅)为Gamma分布。从式(11)可以看出,学生t分布由有限个均值相同、方差不同的高斯分布叠加组成。脉冲压缩后的回波信号已经实现了距离向高分辨,在超分辨过程中通过逐距离单元对方位向回波信号进行处理来提高分辨率。异常值不管出现在某些角度上或随机出现在整个回波信号中,在固定距离单元上异常值的形式是相同的。因此本文引入学生t分布来模拟成像噪声的统计特征。为了简化学生t分布,成像噪声服从
p(n|η,Σ)=N(n|0,(ηΣ)−1) (12) 在式(12)中,假设成像噪声的均值为零。回波信号s服从
p(s|x,η,Σ)=N(s|Hx,(ηΣ)−1) (13) 参数η服从Gamma分布
p(η)=G(η|aη,bη) (14) 其中,aη和bη为超参数,共同控制η的分布。
同时,参数矩阵Σ服从
p(Σ|ν)=M∏m=1G(εm|ν2,ν2) (15) 自由度ν服从Gamma分布
p(ν)=G(ν|aν,bν) (16) 其中,超参数aν和bν共同控制ν的分布。
前视成像中感兴趣目标相对于成像场景是稀疏的,因此本文引入拉普拉斯分层先验分布模拟目标的稀疏特征
p(x|λ)=M∏m=1λm2exp(−λm|xm|) (17) 然后,λ=[λ1,λ2,⋯,λM]T服从
p(λ)=M∏m=1p(λm)=M∏m=1G(λm|aλ,bλ) (18) 其中,超参数aλ和bλ共同控制λ的分布。
上述成像参数的概率模型如图4所示,图中每个结点表示一个变量,链接表示变量之间存在概率关系,箭头表示生成方向,如图中噪声n和目标x共同作用于回波信号s。
3.2 参数估计
3.1节给出了数据异常情况时机载前视成像中噪声和目标的概率模型,本节引入期望最大化方法对成像参数进行估计。期望最大化方法主要分为两步,E步骤和M步骤。E步骤的主要任务是确定目标散射系数x的Q函数。M步骤是通过最大化Q函数来估计目标散射系数x。
E步骤:定义Q函数
Q(ζ;ζ(q))=Eϕ|s,ζ(q)[lnp(s,ζ,ϕ)]=Eη,λ,ν|s,x(q),Σ(q)[lnp(x,η,Σ,λ,ν,s)] (19) 其中,ϕ表示隐藏变量,ϕ={η,λ,ν}。ζ表示模型参数,ζ={x,Σ}。Eϕ|s,ζ(q)[⋅]表示关于p(ϕ|s,ζ(q))的期望。ζ(q)表示第q次迭代中模型参数的值。
M步骤:通过最大化E步骤中Q函数获得模型参数ζ
ζ(q+1)=arg maxζ[Q( ζ;ζ(q))] (20) 通过E步骤,可以得到ν的期望为
⟨ν⟩=2aν+M2bν−M∑m=1(lnεm−εm)−M (21) η的期望为
⟨η⟩=M+2aη(s−Hx)TΣ(s−Hx)+2bη (22) λm的期望为
⟨λm⟩=aλ+1|xm|+bλ (23) 将式(21)、式(22)和式(23)代入式(19),Q函数可写为
Q(ζ;ζ(q))=Eϕ|s,ζ(q)[lnp(s,ζ,ϕ)]=12lnΣ−⟨η⟩2(s−Hx)TΣ(s−Hx)+(⟨ν⟩2−1)lnΣ−⟨ν⟩2Σ−‖Λx‖1+const (24) 其中,Λ=diag{⟨λ1⟩,⟨λ2⟩,⋯,⟨λM⟩}。
在M步骤中,通过求解∇ζQ(ζ;ζ(q))=0找到最优模型参数ζ,可以得到
εm=1+⟨ν⟩⟨η⟩(sm−Hmx)2+⟨ν⟩ (25) x=⟨η⟩(⟨η⟩HTΣH+W)−1HTΣs (26) 其中,Hm为天线方向图矩阵的第m行,W=diag{⟨λ1⟩(|x1|2+δ)−12,⟨λ2⟩(|x2|2+δ)−12,⋯,⟨λM⟩⋅(|xM|2+δ)−12},δ为避免L1范数不可微引入的辅助参数。
3.3 快速参数估计
在目标散射系数x的估计过程中,需要计算矩阵(⟨η⟩HTΣH+W)的逆,其计算复杂度为O(M3),本节引入辅助矩阵并通过矩阵变换对目标散射估计公式进行重写,降低了估计过程的计算复杂度。
方向图矩阵H可以通过奇异值分解写为
H=UAVT (27) 其中,U=[u1,u2,⋯,uM],V=[v1,v2,⋯,vM]为酉矩阵,A=diag{α1,α2,⋯,αM}为H的奇异值矩阵。
在TSVD方法中,目标散射系数x的估计方式为
x=Hr−1s=VrA−1rUTrs=VrVTrx+r∑i=1viα−1iuHin (28) 其中,Hr=UrArVTr是截断的方向图矩阵,r为截断参数。Ur=[u1,u2,⋯,ur],Vr=[v1,v2,⋯,vr]为截断的酉矩阵。Ar=diag{α1,α2,⋯,αr}为截断的奇异值矩阵。
不考虑噪声的情况下,式(28)可以写为
x=VrVTrx (29) 从式(29)近似得到
VrVTr≈I (30) 利用式(30)的性质,可以将式(26)重写为
x=⟨η⟩[⟨η⟩HTΣH+W]−1HTΣs=(HTVrVTrΣ12Σ12VrVTrH+W⟨η⟩−1)−1HTVrVTrΣ12Σ12VrVTrs=(HTVrVTrΣ12Σ12VrVTrH+W⟨η⟩−1)−1HTVrVTrΣ12(Σ12VrVTrHW−1HTVrVTrΣ12+⟨η⟩−1I)×(Σ12VrVTrHW−1HTVrVTrΣ12+⟨η⟩−1I)−1Σ12VrVTrs=W−1HTVrVTrΣ12(Σ12VrVTrHW−1HTVrVTrΣ12+⟨η⟩−1I)−1Σ12VrVTrs=W−1HTVr(VTrHW−1HTVr+⟨η⟩−1VTrΣ−1Vr)−1VTrs=W−1˜HT(˜HW−1˜HT+⟨η⟩−1VTrΣ−1Vr)−1˜s (31) 其中
˜H=VTrH (32) ˜s=VTrs (33) 在原始基于学生t分布的方法中需要计算式(26)矩阵(⟨η⟩HTΣH+W)的逆,矩阵的尺寸为M×M,其计算复杂度为O(M3)。在经过引入截断酉矩阵并通过矩阵变换后,式(31)中矩阵(˜HW−1˜HT+⟨η⟩−1VTrΣ−1Vr)的尺寸为r×r,计算矩阵逆的计算复杂度变为O(r3)。
截断参数r的选择可以参考TSVD方法的思想。按照表1的雷达系统参数,图5给出了方向图矩阵(尺寸400×400)的奇异值曲线,从图中可以看出大奇异值数量为40左右。因为大的奇异值包含了更多的信息,因此本论文中在选择截断参数r时,可以根据奇异值曲线选择拐角处的截断参数,在尽可能保留大奇异值的同时,平衡加速方法的时间性能。
表 1 仿真系统参数Table 1. System parameters of simulation参数 数值 参数 数值 扫描速度(∘/∘ss) 50 载波频率(GHz) 9.5 扫描范围(∘) ±10 信号带宽(MHz) 40 脉冲重复频率(Hz) 1000 平台速度(m/mss) 30 主瓣波束宽度(∘) 3 最后,所提方法的成像参数可以通过式(34)得到
⟨η⟩=M+2aη(s−Hx)TΣ(s−Hx)+2bη⟨λm⟩=aλ+1|xm|+bλ⟨ν⟩=2aν+M2bν−M∑m=1(lnεm−εm)−Mεm=1+⟨ν⟩⟨η⟩(sm−Hmx)2+⟨ν⟩x=W−1˜HT(˜HW−1˜HT+⟨η⟩−1VTrΣ−1Vr)−1˜s (34) 在初始化参数中,本文将待估参数初始化为xm=HTms/(HTmHm),Σ设置为单位矩阵。超参数如aη, bη, aλ, bλ, aν, bν设置为较小的值,如10−4。手动选择合适的截断参数r,在保留大奇异值的同时兼顾算法的加速性能。
3.4 计算复杂度分析
本节对基于学生t分布的方法和其加速方法的计算复杂度进行了分析。假设固定距离单元上回波信号s的尺寸为M×1,迭代次数为q。在分析中,因为基于学生t分布的方法和加速方法中均对η, λ, ν, Σ进行了估计,因此本文忽略了这些参数的计算复杂度,仅考虑计算目标散射系数的计算复杂度。
在基于学生t分布的方法中,计算矩阵HTΣH的复杂度为O(qM3)。计算矩阵(⟨η⟩HTΣH+W)逆的复杂度为O(qM3)。计算⟨η⟩(⟨η⟩HTΣH+W)−1HTΣs的复杂度为O(qM3)。因此总的计算复杂度为O(3qM3)。
在其加速方法中,需要对方向图矩阵H进行奇异值分解,其计算复杂度为O(M3)。计算˜H的复杂度为O(rM2),计算˜s的复杂度为O(rM),计算矩阵˜HW−1˜HT的复杂度为O(qMr2),计算VTrΣ−1Vr的复杂度为O(qMr2)。计算矩阵(˜HW−1˜HT+⟨η⟩−1VTrΣ−1Vr)逆的复杂度为O(qr3)。计算W−1˜HT⋅(˜HW−1˜HT+⟨η⟩−1VTrΣ−1Vr)−1˜s的复杂度为O(qMr2)。加速方法总的计算复杂度为O(3qMr2+qr3+M3+rM2+rM)。
加速方法相比于未加速方法需要进行奇异值分解以及其他附加运算,如计算VTrΣ−1Vr, ˜H, ˜s。当截断参数与回波信号长度接近时,加速方法的计算复杂度会比未加速方法的计算复杂度高。根据前面截断参数的选择,r≪M,因此加速方法的计算复杂度低于基于学生t分布的计算复杂度。
4. 仿真结果
本节将采用仿真数据和半实测数据验证所提方法的有效性。本文将所提加速的基于学生t分布方法与Tikhonov正则化(Tikhonov regularization, REGU)方法,LRIAA方法,基于学生t分布的方法进行了比较。为方便起见,将基于学生t分布的方法简写为MBSD (Method Based on Student-t Distribution),加速的基于学生t分布的方法简写为AMBSD (Accelerated Method Based on Student-t Distribution)。在仿真中,假设信噪比(Signal-to-Noise Ratio, SNR)为信号与高斯白噪声的比值。仿真中的异常值为生成回波信号后随机加入。
4.1 点目标仿真
图6(a)为原始点目标场景,其中两个目标分别位于−0.5∘和0.5∘,雷达仿真系统参数如表1所示。
图6(b)为信噪比为15 dB的雷达回波信号,回波信号的尺寸为400×1。在回波信号中,我们加入了5个幅值随机的异常值。从图中可以看出,经过雷达波束的调制,两个目标被混合在一起无法分辨,并且有异常值融合在了两个目标之间。图6(c)—图6(f)分别为不同方法的处理结果。在图6(c)中,REGU方法可以将两个目标大致区分开,但两目标之间的谷值还是较高。而且受异常值影响,很难判断目标的真实位置。图6中LRIAA方法可以很好地区分目标,但是同REGU方法一样,因为没有考虑异常值对前视成像的影响,一部分的异常值被重构为虚假目标。图6(e)和图6(f)分别为MBSD方法以及AMBSD方法的处理结果,其中截断参数设为50。从图中可以看出,MBSD方法和AMBSD方法均可以将两相邻目标进行区分并且可以抑制回波中的异常值。
4.2 MSE和计算时间对比
为了评估所提方法的优越性,本文引入均方误差(Mean Square Error, MSE)来定量地观察不同方法的性能。MSE可以通过式(35)得到
MSE=1Mc(1M‖x−ˆx‖22) (35) 其中,x是真实的目标值,ˆx是x的估计值,Mc是蒙特卡罗实验的次数。仿真分别在信噪比为30 dB, 25 dB, 20 dB, 15 dB和10 dB的环境下进行500次蒙特卡罗实验。MSE曲线如图7所示,可以观察到随着信噪比的提升,不同方法的性能均得到提升。相对于其他方法,MBSD方法和AMBSD方法可以获得更好的MSE结果。
然后我们对不同方法的计算时间进行了对比。仿真时的CPU为 Intel Core i5-12500H 3.10 GHz,RAM为16 GB。仿真软件为MATLAB 2021a。计算时间对比结果如图8所示。从图中可以看出REGU方法的计算时间相对于其他方法更低,但其角度分辨率较低。LRIAA方法的运行时间低于MBSD方法,但其MSE结果不如MBSD方法。反观AMBSD方法(r=50)在重构目标准确性和运算时间上都拥有一个良好的性能。
4.3 截断参数对性能的影响
基于点目标仿真,本节仿真了20 dB信噪比时不同截断参数情况下MSE和计算时间。截断参数对MSE影响的仿真结果如图9所示。从图中可以看出,随着截断参数r取值的逐渐增大,AMBSD方法的MSE逐渐下降并趋于MBSD方法的MSE。并且可以看出MSE曲线的趋势与图5中奇异值曲线的趋势类似,说明大奇异值保留了大部分的原始信息,对目标参数的重构误差影响较大。截断参数对计算时间的影响情况如图10所示。从图中可以看出当截断参数较大时,因为存在附加运算,因此加速方法的运行时间比未加速方法的运行时间长。当截断参数较小时,AMBSD方法的计算时间比MBSD方法的计算时间更短,并且截断参数与方位向回波尺寸的比值越小,AMBSD方法相比于MBSD方法的加速效果更加明显。
4.4 面目标仿真结果
本节我们通过面目标仿真验证所提方法的有效性。图11为原始目标场景,尺寸为300×400(距离向×方位向),图中8个合成目标分别分布于不同位置。AMBSD方法的截断参数设为50。表2给出了面目标仿真的雷达系统参数。受到电磁干扰或雷达设备性能影响,异常值可能会出现在回波中的某个方位角上或者随机出现在整个回波上,因此我们将面目标仿真分为两部分。
表 2 面目标仿真系统参数Table 2. System parameters of area target simulation参数 数值 参数 数值 扫描速度(∘/∘ss) 50 载波频率(GHz) 9.5 扫描范围(∘) ±10 信号带宽(MHz) 40 脉冲重复频率(Hz) 1000 平台速度(m/mss) 30 主瓣波束宽度(∘) 5 (1) 受电磁干扰时前视成像:图12(a)为雷达回波信号,此时的信噪比为10 dB,有8个随机幅值的异常值随机分布在某些方位角上。从回波信号中可以看出目标被混合在一起,无法分辨。图12(b)为LRIAA方法的处理结果,因为没有考虑异常值,LRIAA方法受到异常值的影响,2500 m附近的重构目标被干扰严重。在图12(c)和图12(d)中,MBSD方法和AMBSD方法可以很好地抑制异常值并且区分目标。为了更好地量化重构结果,表3给出了MSE结果对比和运行时间对比。从表中可以看出,所提AMBSD方法可以用更短的运行时间获得更好的重构结果。
表 3 受电磁干扰时面目标仿真的MSE和运行时间Table 3. MSE and running time of area target simulation with electromagnetic interference方法 MSE 运行时间(s) LRIAA方法 6.15×10–3 4.90 MBSD方法 0.65×10–3 23.14 AMBSD方法 0.70×10–3 3.82 (2) 设备性能异常时前视成像:从表4的运行时间可以看出,所提AMBSD方法具有更快的运行时间。图13(a)为雷达回波信号,此时的信噪比为10 dB,异常值被随机地添加到回波信号中。图13(b)为LRIAA方法处理结果,图中目标可以被分开,但是异常值依旧零散地分布于整个重构场景中。相比于LRIAA方法的处理结果,MBSD方法和AMBSD方法获得了更好的重构结果。
表 4 设备性能异常时面目标仿真的MSE和运行时间Table 4. MSE and running time of area target simulation with abnormal equipment performance方法 MSE 运行时间(s) LRIAA方法 1.09×10–3 4.80 MBSD方法 0.80×10–3 23.03 AMBSD方法 0.81×10–3 3.90 4.5 半实测数据处理结果
前面的仿真实验证明了所提AMBSD方法可以用更快的速度抑制回波信号中的异常值,提高前视成像的角度分辨率。本节采用半实测数据进一步验证所提方法的性能。图14(a)为机载雷达实测回波信号,其中红色矩形框内存在两个混合在一起的目标,回波信号尺寸为513×208(距离向×方位向)。AMBSD方法的截断参数设为50。图14(b)和图14(c)为MBSD方法和AMBSD方法的处理结果。从处理结果中可以看出,各方法可以对红色矩形框中的目标进行分辨,提高前视成像的角度分辨率,证明了MBSD方法和AMBSD方法对实测数据的有效性。
(1) 受电磁干扰时前视成像:图15(a)为半实测回波信号,我们将异常值随机添加到不同的方位角上。图15(b)—图15(d)为不同方法的处理结果。从图15(b)中可以看出,LRIAA方法对异常值的抑制能力较差,并且红色矩形框中的目标没有被区分开。相比于LRIAA方法,MBSD方法和AMBSD方法均可以抑制异常值,并且红色矩形框中的目标也可以被区分出来。表5给出了不同方法的运行时间对比,可以看出AMBSD方法具有更短的运行时间。
表 5 半实测数据运行时间(s)Table 5. Running time of semi-real data (s)方法 受电磁干扰时运行时间 设备性能异常时运行时间 LRIAA方法 1.98 1.89 MBSD方法 4.57 4.58 AMBSD方法 1.18 1.19 (2) 设备性能异常时前视成像:图16(a)为回波信号,此时异常值随机分布于回波信号中。图16(b)—图16(d)为不同方法的处理结果。从图16(c)和图16(d)中可以看出MBSD方法和AMBSD方法可以抑制异常值并且区分红色矩形框中的目标。并且表5说明AMBSD方法可以用更短的运行时间获得与MBSD方法近似的分辨性能。相比于表3和表4中AMBSD方法与MBSD方法的加速比值,因为在半实测仿真中截断参数与方位向尺寸的比值增大,因此表5中的加速比值更小。
5. 结语
本文提出了一种机载雷达超分辨方法实现回波数据异常时快速前视成像。该方法通过利用学生t分布对异常值的鲁棒性,使噪声模型对存在异常值时的回波信号更加灵活。为了更好地重构目标,引入拉普拉斯分层先验来模拟目标的稀疏特征。然后基于期望最大化方法对成像参数进行估计。为了降低参数估计过程的运算量,本文将截断后的酉矩阵引入到目标散射系数的估计公式中,并通过矩阵变换的方法将原方法中需要对M×M维矩阵求逆转变为对r×r维矩阵求逆。从仿真数据和半实测数据中可以看出,本文所提出的快速方法可以用更短的时间实现在回波数据异常情况下的前视超分辨成像。
-
图 2 Performance comparison of common polarimetric optimization detection methods[18]
图 3 PWF processing results with different textural distributions[21]
图 4 Dense ship detection in PolSAR images based on the adaptive truncation method[28]
图 5 Performance comparison of PNF and NPNF[8]
图 6 Comparison of detection results among different methods for small targets[36]
图 7 Comparison results among different ship detectors with SPN[51]
图 8 Performance comparision among different ship detection methods with NPCM[6]
图 9 Moving target detection results in different polarimetric channels in PolSAR images[69]
图 10 Simulation results of polarimetric along-track interferometry moving target detection[76]
图 11 Wake detection results by PWF and Radon transform[90]
图 12 Ship wake detection results of low-rank and sparse decomposition combined with polarization[98]
图 13 P2P-CNN network structure and detection performance[107]
表 1 PolSAR图像舰船目标检测方法适用场景和优缺点总结
Table 1. Applicable scenarios and pros & cons summary of PolSAR ship detection methods
检测分类 适用场景 具体检测方法 优点 缺点 目标极化特征检测 主要适用于舰船目标本体极化特征和杂波特征有一定差异情形下的检测 简单极化合成检测技术 早期PolSAR目标检测方法,简单易实现且效果优于单极化SAR图像目标检测方法 一定程度上利用了幅度或相位信息,对极化信息的利用不够充分 基于极化最优化的检测技术 1. 充分利用了各极化通道相关信息
2. 通过对PolSAR图像进行优化,达到最佳对比效果要根据实际情况选择不同的极化优化准则,同时滑窗的选择也对性能影响较大 基于散射机理的检测技术 具有较强的物理可解释性 对目标类型和海况状态的适应性需要提高 基于空间邻域的检测技术 融合空域与极化信息,大幅增强目标检测能力 后期数据处理需要降维以避免维数灾难 慢速运动目标检测 主要适用于目标杂波极化特征差异较小但有一定速度差异的情形,同时也能提取目标运动信息以获取实时海面态势 单通道慢动目标检测 利用交叉极化信息增强慢动目标的检测效果 静止目标模糊抑制和慢动目标检测难以同时实现 虚拟干涉慢动目标检测 解决了实际干涉数据来源缺乏的问题 子孔径分解的个数与重叠度的选择较为困难 多通道慢动目标检测 融合多平台和极化信息的优势,极大增强运动目标检测能力 复杂结构和成本限制工程应用 舰船目标尾迹检测 不是检测舰船目标本体,而是检测运动产生的尾迹,主要针对小目标和
隐身目标尾迹边缘线性特征检测 检测问题抽象成线形特征检测问题,方法相对简单 杂波背景下不同类型弱尾迹的检测困难 尾迹区域背景差异检测 尾迹区域和背景海面由于散射机理不同,在频谱特性、统计特征、极化特性等方面均存在较大差异,效果好 理论分析和方法实现更加复杂 基于深度学习的目标检测 不需要人工提取目标特征,在PolSAR图像智能解译中取得了巨大成功 基于卷积神经网络的检测方法 结构灵活、能够自动提取结构化特征,不仅能提取图像的低维特征,而且能提取图像的高维特征 1. 需要大量样本进行训练
2. 可解释性有待研究表 1 Applicable scenarios and pros & cons summary of PolSAR ship detection methods
Detection method Advantage Disadvantage Target polarimetric characteristic detection method (mainly applied to the detection of ship and clutter when the polarimetric characteri
stics are different)Simple polarimetric channel
synthesis technologyThe early PolSAR target dete
ction method is simple and easy to realize, and the effect is better than the single PolSAR image target detection methodTo some extent, the amplitude or phase information is used, but the polarimetric information is not fully utilized Polarimetric optimization
detection technology1. The information of each polari
metric channel is fully utilized
2. PolSAR images are optimized to achieve the best contrast effectDifferent optimization criteria should be selected according to the actual situation, and the selection of a sliding window also has a significant influence on the performance Detection technology based on the
scattering mechanismIt has strong physical interpre
tabilityAdaptability to target types and sea state needs to be improved Spatial neighborhood detection
methodThe fusion of spatial and polarime
tric information considerably enhances the target detection capabilityDimensions need to be reduced to avoid a dimensional disaster Slow-moving target detection method (mainly applied to the situation where the polarimetric characteristic difference between target and clutter is small but thesre is a certain velocity diffe
rence; simultaneously, it can also extract the target motion informa
tion to obtain the real-time sea surface situation)Single-channel PolSAR-GMTI Cross-polarimetric information is used to enhance the detection effect of slow-moving targets Suppression of stationary targets and detection of slow-moving targets are difficult to achieve simultaneously Virtual interferometric
slow-moving target detectionThe problem of the lack of actual interference data source is solved It is difficult to choose the number of subaperture decompo
sition and overlap degreeMultichannel PolSAR-GMTI Integrating the advantages of multi-platform and polarization information considerably enhan
ces the ability to move target detectionComplex structure and cost limit engineering applications Ship wake detection method (it is used not to detect the ship target body but to detect the wake gene
rated by the movement, mainly for small targets and stealth targets)Linear feature detection of
wake edgeThe detection problem is abstrac
ted into the linear feature detec
tion problem, and the method is relatively simpleIt is difficult to detect different types of weak wakes in the cluttered background Background difference detection
in the wake regionDue to the different scattering mechanisms of the wake region and background sea surface, there are considerable differences in spectrum characteristics, statisti
cal characteristics, polarimetric characteristics, and other aspects, and the effect is goodThe theoretical analysis and method implementation are more complex Target detection algorithm based on deep learning (it has achieved great success in intelligent PolSAR image interpretation without extracting target features man
ually)Detection method based on CNN Flexible structure can automa
tically extract structured features,
including both low-dimensional and high-dimensional features of images1. It needs a large number of samples to train with
2. Interpretability needs to be investigated -
[1] CRISP D J. The state-of-the-art in ship detection in synthetic aperture radar imagery[R]. DSTO-RR-0272, 2004. [2] RANSON K J and SUN Guoqing. An evaluation of AIRSAR and SIR-C/X-SAR images for mapping northern forest attributes in Maine, USA[J]. Remote Sensing of Environment, 1997, 59(2): 203–222. doi: 10.1016/S0034-4257(96)00154-X [3] 代大海, 王雪松, 肖顺平, 等. PolSAR系统与技术的发展趋势[J]. 雷达科学与技术, 2008, 6(1): 15–22. doi: 10.3969/j.issn.1672-2337.2008.01.003DAI Dahai, WANG Xuesong, and XIAO Shunping, et al. Development trend of PolSAR system and technology[J]. Radar Science and Technology, 2008, 6(1): 15–22. doi: 10.3969/j.issn.1672-2337.2008.01.003 [4] 匡纲要, 陈强, 蒋咏梅, 等. 极化合成孔径雷达基础理论及其应用[M]. 长沙: 国防科技大学出版社, 2011.KUANG Gangyao, CHEN Qiang, JIANG Yongmei, et al. Polarimetric Synthetic Aperture Radar Basic Principles and Its Applications[M]. Changsha: University of Defense Science and Technology Press, 2011. [5] 刘涛, 崔浩贵, 谢恺, 等. 极化合成孔径雷达图像解译技术[M]. 北京: 国防工业出版社, 2017.LIU Tao, CUI Haogui, XIE Kai, et al. Interpretation Techniques in Polarimetic Synthetic Apture Radar Imagery[M]. Beijing: National Defense Industry Press, 2017. [6] MARINO A, CLOUDE S R, and WOODHOUSE I H. Detecting depolarized targets using a new geometrical perturbation filter[J]. IEEE Transactions on Geoscience and Remote Sensing, 2012, 50(10): 3787–3799. doi: 10.1109/TGRS.2012.2185703 [7] LIU Tao, YANG Ziyuan, MARINO A, et al. PolSAR ship detection based on neighborhood polarimetric covariance matrix[J]. IEEE Transactions on Geoscience and Remote Sensing, 2020, in press. doi: 10.1109/TGRS.2020.3022181 [8] LIU Tao, YANG Ziyuan, ZHANG Tao, et al. A new form of the polarimetric notch filter[J]. IEEE Geoscience and Remote Sensing Letters, 2020, in press. doi: 10.1109/LGRS.2020.3020052 [9] SCIOTTI M, PASTINA D, and LOMBARDO P. Polarimetric detectors of extended targets for ship detection in SAR images[C]. IEEE 2001 International Geoscience and Remote Sensing Symposium Scanning the Present and Resolving the Future, Sydney, Australia, 2001. [10] 张鹏, 张嘉峰, 刘涛. 雷达多视极化检测器性能对比分析[J]. 电波科学学报, 2017, 32(4): 416–426. doi: 10.13443/j.cjors.2017022401ZHANG Peng, ZHANG Jiafeng, and LIU Tao. Contrastive analysis of the performances of radar multi-look polarimetric detectors[J]. Chinese Journal of Radio Science, 2017, 32(4): 416–426. doi: 10.13443/j.cjors.2017022401 [11] BOERNER W M, KOSTINSKI A B, and JAMES B D. On the concept of the polarimetric matched filter in high resolution radar imaging: An alternative for speckle reduction[C]. International Geoscience and Remote Sensing Symposium, ‘Remote Sensing: Moving Toward the 21st Century’, Edinburgh, UK, 1988: 69–72. [12] NOVAK L M, SECHTIN M B, and CARDULLO M J. Studies of target detection algorithms that use polarimetric radar data[J]. IEEE Transactions on Aerospace and Electronic Systems, 1989, 25(2): 150–165. doi: 10.1109/7.18677 [13] CHANEY R D, BUD M C, and NOVAK L M. On the performance of polarimetric target detection algorithms[J]. IEEE Aerospace and Electronic Systems Magazine, 1990, 5(11): 10–15. doi: 10.1109/62.63157 [14] NOVAK L M and BURL M C. Optimal speckle reduction in polarimetric SAR imagery[J]. IEEE Transactions on Aerospace and Electronic Systems, 1990, 26(2): 293–305. doi: 10.1109/7.53442 [15] YANG Jian, YAMAGUCHI Y, BOERNER W M, et al. Numerical methods for solving the optimal problem of contrast enhancement[J]. IEEE Transactions on Geoscience and Remote Sensing, 2000, 38(2): 965–971. doi: 10.1109/36.84197 [16] YANG Dongwen, DU Lan, LIU Hongwei, et al. Novel polarimetric contrast enhancement method based on minimal clutter to signal ratio subspace[J]. IEEE Transactions on Geoscience and Remote Sensing, 2019, 57(11): 8570–8583. doi: 10.1109/TGRS.2019.2921629 [17] NOVAK L M. Target detection studies using fully polarimetric data collected by the Lincoln Laboratory MMW SAR[C]. 92 International Conference on Radar, Brighton, UK, 1992. [18] LIU Tao, YANG Ziyuan, YANG Jian, et al. CFAR ship detection methods using compact polarimetric SAR in a K-Wishart distribution[J]. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2019, 12(10): 3737–3745. doi: 10.1109/JSTARS.2019.2923009 [19] LOPES A and SERY F. Optimal speckle reduction for the product model in multilook polarimetric SAR imagery and the Wishart distribution[J]. IEEE Transactions on Geoscience and Remote Sensing, 1997, 35(3): 632–647. doi: 10.1109/36.581979 [20] LIU Guoqing, HUANG Shunji, TORRE A, et al. The multilook polarimetric whitening filter (MPWF) for intensity speckle reduction in polarimetric SAR images[J]. IEEE Transactions on Geoscience and Remote Sensing, 1998, 36(3): 1016–1020. doi: 10.1109/36.673694 [21] LIU Tao, ZHANG Jiafeng, GAO Gui, et al. CFAR ship detection in polarimetric synthetic aperture radar images based on whitening filter[J]. IEEE Transactions on Geoscience and Remote Sensing, 2019, 58(1): 58–81. doi: 10.1109/TGRS.2019.2931353 [22] GAO Gui, LI Gaosheng, and LI Yipeng. Shape parameter estimator of the generalized Gaussian distribution based on the MoLC[J]. IEEE Geoscience and Remote Sensing Letters, 2018, 15(3): 350–354. doi: 10.1109/LGRS.2017.2787558 [23] GAO Gui, OUYANG Kewei, LUO Yongbo, et al. Scheme of parameter estimation for generalized gamma distribution and its application to ship detection in SAR images[J]. IEEE Transactions on Geoscience and Remote Sensing, 2017, 55(3): 1812–1832. doi: 10.1109/TGRS.2016.2634862 [24] GAO Gui, LUO Yongbo, OUYANG Kewei, et al. Statistical modeling of PMA detector for ship detection in high-resolution dual-polarization SAR images[J]. IEEE Transactions on Geoscience and Remote Sensing, 2016, 54(7): 4302–4313. doi: 10.1109/TGRS.2016.2539200 [25] GAO Gui, WANG Xiaoyang, and NIU Min. Statistical modeling of the reflection symmetry metric for sea clutter in dual-polarimetric SAR data[J]. IEEE Journal of Oceanic Engineering, 2016, 41(2): 339–345. doi: 10.1109/JOE.2015.2458231 [26] 冷祥光, 计科峰, 熊博莅, 等. 面向舰船目标检测的单通道复值SAR图像统计建模方法研究[J]. 雷达学报, 2020, 9(3): 477–496. doi: 10.12000/JR20070LENG Xiangguang, JI Kefeng, XIONG Boli, et al. Statistical modeling methods of single-channel complex-valued SAR images for ship detection[J]. Journal of Radars, 2020, 9(3): 477–496. doi: 10.12000/JR20070 [27] TAO Ding, ANFINSEN S N, and BREKKE C. Robust CFAR detector based on truncated statistics in multiple-target situations[J]. IEEE Transactions on Geoscience and Remote Sensing, 2016, 54(1): 117–134. doi: 10.1109/TGRS.2015.2451311 [28] LIU Tao, YANG Ziyuan, MARINO A, et al. Robust CFAR detector based on truncated statistics for polarimetric synthetic aperture radar[J]. IEEE Transactions on Geoscience and Remote Sensing, 2020, 58(9): 6731–6747. doi: 10.1109/TGRS.2020.2979252 [29] LANG Haitao, XI Yuyang, and ZHANG Xi. Ship detection in high-resolution SAR images by clustering spatially enhanced pixel descriptor[J]. IEEE Transactions on Geoscience and Remote Sensing, 2019, 57(8): 5407–5423. doi: 10.1109/TGRS.2019.2899337 [30] NUNZIATA F, MIGLIACCIO M, and BROWN C E. Reflection symmetry for polarimetric observation of man-made metallic targets at sea[J]. IEEE Journal of Oceanic Engineering, 2012, 37(3): 384–394. doi: 10.1109/JOE.2012.2198931 [31] WANG Na, SHI Gongtao, LIU Li, et al. Polarimetric SAR target detection using the reflection symmetry[J]. IEEE Geoscience and Remote Sensing Letters, 2012, 9(6): 1104–1108. doi: 10.1109/LGRS.2012.2189548 [32] MARINO A. A notch filter for ship detection with polarimetric SAR data[J]. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2013, 6(3): 1219–1232. doi: 10.1109/JSTARS.2013.2247741 [33] MARINO A and HAJNSEK I. Statistical tests for a ship detector based on the polarimetric notch filter[J]. IEEE Transactions on Geoscience and Remote Sensing, 2015, 53(8): 4578–4595. doi: 10.1109/TGRS.2015.2402312 [34] GAO Gui and SHI Gongtao. CFAR ship detection in nonhomogeneous sea clutter using polarimetric SAR data based on the notch filter[J]. IEEE Transactions on Geoscience and Remote Sensing, 2017, 55(8): 4811–4824. doi: 10.1109/TGRS.2017.2701813 [35] GAO Gui and SHI Gongtao. Ship detection in dual-channel ATI-SAR based on the notch filter[J]. IEEE Transactions on Geoscience and Remote Sensing, 2017, 55(8): 4795–4810. doi: 10.1109/TGRS.2017.2701810 [36] GAO Gui, GAO Sheng, HE Juan, et al. Ship detection using compact polarimetric SAR based on the notch filter[J]. IEEE Transactions on Geoscience and Remote Sensing, 2018, 56(9): 5380–5393. doi: 10.1109/TGRS.2018.2815582 [37] CLOUDE S R and POTTIER E. A review of target decomposition theorems in radar polarimetry[J]. IEEE Transactions on Geoscience and Remote Sensing, 1996, 34(2): 498–518. doi: 10.1109/36.485127 [38] RINGROSE R and HARRIS N. Ship detection using polarimetric SAR data[J]. European Space Agency ESA SP, 2000, 450(450): 687. [39] TOUZI R, CHARBONNEAU F, HAWKINS R K, et al. Ship-sea contrast optimization when using polarimetric SARs[C]. IEEE 2001 International Geoscience and Remote Sensing Symposium Scanning the Present and Resolving the Future, Sydney, Australia, 2001. [40] TOUZI R. Calibrated polarimetric SAR data for ship detection[C]. IEEE 2000 International Geoscience and Remote Sensing Symposium. Taking the Pulse of the Planet: The Role of Remote Sensing in Managing the Environment. Proceedings (Cat. No. 00CH37120), Honolulu, USA, 2000. [41] CHEN Jiong, CHEN Yilun, and YANG Jian. Ship detection using polarization cross-entropy[J]. IEEE Geoscience and Remote Sensing Letters, 2009, 6(4): 723–727. doi: 10.1109/LGRS.2009.2024224 [42] SUGIMOTO M, OUCHI K, and NAKAMURA Y. On the novel use of model-based decomposition in SAR polarimetry for target detection on the sea[J]. Remote Sensing Letters, 2013, 4(9): 843–852. doi: 10.1080/2150704X.2013.804220 [43] YANG Jian, DONG Guiwei, PENG Yingning, et al. Generalized optimization of polarimetric contrast enhancement[C]. IEEE Antennas and Propagation Society International Symposium. Digest. Held in conjunction with: USNC/CNC/URSI North American Radio Sci. Meeting (Cat. No. 03CH37450), Columbus, USA, 2003. [44] YIN Junjun, YANG Jian, XIE Chunhua, et al. An improved generalized optimization of polarimetric contrast enhancement and its application to ship detection[J]. IEICE Transactions on Communications, 2013, E96.B(7): 2005–2013. doi: 10.1587/transcom.E96.B.2005 [45] TOUZI R, HURLEY J, and VACHON P W. Optimization of the degree of polarization for enhanced ship detection using polarimetric RADARSAT-2[J]. IEEE Transactions on Geoscience and Remote Sensing, 2015, 53(10): 5403–5424. doi: 10.1109/TGRS.2015.2422134 [46] BORDBARI R and MAGHSOUDI Y. A new target detector based on subspace projections using polarimetric SAR data[J]. IEEE Transactions on Geoscience and Remote Sensing, 2019, 57(5): 3025–3039. doi: 10.1109/TGRS.2018.2879681 [47] 殷君君, 安文韬, 杨健. 基于极化散射参数与Fisher-OPCE的监督目标分类[J]. 清华大学学报: 自然科学版, 2011, 51(12): 1782–1786.YIN Junjun, AN Wentao, and YANG Jian. Supervised target classification using polarimetric scattering parameters and Fisher-OPCE[J]. Journal of Tsinghua University:Science and Technology, 2011, 51(12): 1782–1786. [48] CUI Xingchao, CHEN Siwei, and SU Yi. Ship detection in polarimetric SAR image based on similarity test[C]. 2019 IEEE International Geoscience and Remote Sensing Symposium, Yokohama, Japan, 2019. doi: 10.1109/IGARSS.2019.8900480. [49] HE Jinglu, WANG Yinghua, LIU Hongwei, et al. A novel automatic PolSAR ship detection method based on superpixel-level local information measurement[J]. IEEE Geoscience and Remote Sensing Letters, 2018, 15(3): 384–388. doi: 10.1109/LGRS.2017.2789204 [50] LANG Haitao, TAO Yunhong, NIU Lihui, et al. A new scattering similarity based metric for ship detection in polarimetric synthetic aperture radar image[J]. Acta Oceanologica Sinica, 2020, 39(5): 145–150. doi: 10.1007/s13131-020-1563-7 [51] CUI Xingchao, SU Yi, and CHEN Siwei. A saliency detector for polarimetric SAR ship detection using similarity test[J]. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2019, 12(9): 3423–3433. doi: 10.1109/JSTARS.2019.2925833 [52] HUANG Xiaojing, HUANG Pingping, DONG Lixia, et al. Saliency detection based on distance between patches in polarimetric SAR images[C]. 2014 IEEE Geoscience and Remote Sensing Symposium, Quebec City, Canada, 2014. [53] JÄGER M and HELLWICH O. Saliency and salient region detection in SAR polarimetry[C]. 2005 IEEE International Geoscience and Remote Sensing Symposium, Seoul, South Korea, 2005. [54] WANG Haipeng, XU Feng, and CHEN Shanshan. Saliency detector for SAR images based on pattern recurrence[J].IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2016, 9(7): 2891–2900. doi: 10.1109/JSTARS.2016.2521709 [55] LIN Huiping, CHEN Hang, JIN Kan, et al. Ship detection with superpixel-level fisher vector in high-resolution SAR images[J]. IEEE Geoscience and Remote Sensing Letters, 2020, 17(2): 247–251. doi: 10.1109/LGRS.2019.2920668 [56] ZHANG Tao, JI Jinsheng, LI Xiaofeng, et al. Ship detection from PolSAR imagery using the complete polarimetric covariance difference matrix[J]. IEEE Transactions on Geoscience and Remote Sensing, 2018, 57(5): 2824–2839. doi: 10.1109/TGRS.2018.2877821 [57] ZHANG Tao, YANG Zhen, GAN Hongping, et al. PolSAR ship detection using the joint polarimetric information[J]. IEEE Transactions on Geoscience and Remote Sensing, 2020, 58(11): 8225–8241. doi: 10.1109/TGRS.2020.2989425 [58] PERRY R P, DIPIETRO R C, and FANTE R L. SAR imaging of moving targets[J]. IEEE Transactions on Aerospace and Electronic Systems, 1999, 35(1): 188–200. doi: 10.1109/7.745691 [59] GAO Gui, WANG Xiaoyang, and LAI Tao. Detection of moving ships based on a combination of magnitude and phase in along-track interferometric SAR—Part II: Statistical modeling and CFAR detection[J]. IEEE Transactions on Geoscience and Remote Sensing, 2015, 53(7): 3582–3599. doi: 10.1109/TGRS.2014.2379351 [60] GAO Gui, WANG Xiaoyang, and LAI Tao. Detection of moving ships based on a combination of magnitude and phase in along-track interferometric SAR—Part I: SIMP metric and its performance[J]. IEEE Transactions on Geoscience and Remote Sensing, 2015, 53(7): 3565–3581. doi: 10.1109/TGRS.2014.2379352 [61] GAO Gui and SHI Gongtao. The CFAR detection of ground moving targets based on a joint metric of SAR interferogram’s magnitude and phase[J]. IEEE Transactions on Geoscience and Remote Sensing, 2012, 50(9): 3618–3624. doi: 10.1109/TGRS.2012.2184836 [62] PARK H R, LI Jian, and WANG Hong. Polarization-space-time domain generalized likelihood ratio detection of radar targets[J]. Signal Processing, 1995, 41(2): 153–164. doi: 10.1016/0165-1684(94)00097-J [63] PARK H R, KWAG Y K, and WANG Hong. An efficient adaptive polarimetric processor with an embedded CFAR[J]. ETRI Journal, 2003, 25(3): 171–178. doi: 10.4218/etrij.03.0102.0316 [64] WANG Genyuan, XIA Xianggen, and CHEN V C. Radar imaging of moving targets in foliage using multifrequency multiaperture polarimetric SAR[J]. IEEE Transactions on Geoscience and Remote Sensing, 2003, 41(8): 1755–1764. doi: 10.1109/TGRS.2003.813501 [65] FRIEDLANDER B and PORAT B. VSAR: A high resolution radar system for detection of moving targets[J]. IEE Proceedings-Radar,Sonar and Navigation, 1997, 144(4): 205–218. doi: 10.1049/ip-rsn:19971309 [66] LIU Zhongxun, DAI Dahai, LI Dun, et al. Optimal polarimetric interferometry coherence analysis in detection and location of moving target with SAR[C]. The 1st Asian and Pacific Conference on Synthetic Aperture Radar, Huangshan, China, 2007. [67] MATTAR K E, LIU Chen, and SABRY R. Polarimetric SAR interferometry: Investigations using EC CV-580 SAR data[R]. Ottawa: Defence Research and Development Canada Ottawa, 2005. [68] ZOU Bin, WEI Tao, and ZHANG Lamei. Moving targets detection and analysis on multi-look polarimetric SAR images using PWF method[C]. 2008 IEEE International Geoscience and Remote Sensing Symposium, Boston, USA, 2008. [69] LIU Chen and GIERULL C H. A new application for polsar imagery in the field of moving target indication/ship detection[J]. IEEE Transactions on Geoscience and Remote Sensing, 2007, 45(11): 3426–3436. doi: 10.1109/TGRS.2007.907192 [70] LIU Chen. Time-frequency analysis of PolSAR moving target data[J]. Canadian Journal of Remote Sensing, 2007, 33(4): 237–249. doi: 10.5589/m07-019 [71] STACY N and PREISS M. Polarimetric ATI slow target detection in a log likelihood framework[C]. 2013 IEEE International Geoscience and Remote Sensing Symposium, Melbourne, Australia, 2013. [72] 李延伟. 单/双基地极化干涉SAR信号建模、检测及参数反演方法研究[D]. [博士论文], 国防科学技术大学, 2010.LI Yanwei. Study on the method of the modeling, the detection and the parameter inversion for the mono/bi static polarimetric SAR interferometry[D]. [Ph. D. dissertation], National University of Defense Technology, 2010. [73] CHIU S, GIERULL C, and RASHID M. First results of experimental polarimetric SAR-GMTI modes on RADARSAT-2[C]. 2018 IEEE International Geoscience and Remote Sensing Symposium, Valencia, Spain, 2018. [74] CHIU S, GIERULL C, and RASHID M. Ship detection, discrimination, and motion estimation via spaceborne polarimetric SAR-GMTI[C]. 2019 IEEE Radar Conference, Boston, USA, 2019. [75] ZHANG Peng, ZHANG Jiafeng, and LIU Tao. Constant false alarm rate detection of slow targets in polarimetric along-track interferometric synthetic aperture radar imagery[J]. IET Radar,Sonar&Navigation, 2019, 13(1): 31–44. doi: 10.1049/iet-rsn.2018.5082 [76] 张鹏, 张嘉峰, 刘涛. 基于相干度优化的极化顺轨干涉SAR慢小目标CFAR检测[J]. 北京航空航天大学学报, 2019, 45(3): 575–587. doi: 10.13700/j.bh.1001-5965.2018.0322ZHANG Peng, ZHANG Jiafeng, and LIU Tao. Slow and small target CFAR detection of polarimetric along-track interferometric SAR using coherence optimization[J]. Journal of Beijing University of Aeronautics and Astronautics, 2019, 45(3): 575–587. doi: 10.13700/j.bh.1001-5965.2018.0322 [77] OUCHI K. On the multilook images of moving targets by synthetic aperture radars[J]. IEEE Transactions on Antennas and Propagation, 1985, 33(8): 823–827. doi: 10.1109/TAP.1985.1143684 [78] KIRSCHT M. Detection and velocity estimation of moving objects in a sequence of single-look SAR images[C]. 1996 International Geoscience and Remote Sensing Symposium, Lincoln, USA, 1996. [79] GAO Fei, MAO Shiyi, SUN Jinping, et al. A tri-look detector for single channel SAR-GMTI[C]. 2009 IET International Radar Conference, Guilin, China, 2009. [80] 康雪艳, 杨汝良. 利用DPCA方法对机载单天线SAR实际数据进行动目标检测[J]. 遥感技术与应用, 2004, 19(3): 182–186. doi: 10.3969/j.issn.1004-0323.2004.03.009KANG Xueyan and YANG Ruliang. Moving targets detection using DPCA Method for single-antenna airborne SAR real data[J]. Remote Sensing Technology and Application, 2004, 19(3): 182–186. doi: 10.3969/j.issn.1004-0323.2004.03.009 [81] 张露, 郭华东, 韩春明, 等. 基于子孔径分解的SAR动目标检测方法[J]. 电子学报, 2008, 36(6): 1210–1213. doi: 10.3321/j.issn:0372-2112.2008.06.034ZHANG Lu, GUO Huadong, HAN Chunming, et al. Moving targets detection in SAR images based on sub-aperture decomposition[J]. Acta Electronica Sinica, 2008, 36(6): 1210–1213. doi: 10.3321/j.issn:0372-2112.2008.06.034 [82] 康雪艳, 杨汝良. 对机载单天线SAR实际数据进行ATI动目标检测的新方法[J]. 电子学报, 2005, 33(3): 416–418. doi: 10.3321/j.issn:0372-2112.2005.03.008KANG Xueyan and YANG Ruliang. A new method of moving target detection using ATI for single-antenna airborne SAR real data[J]. Acta Electronica Sinica, 2005, 33(3): 416–418. doi: 10.3321/j.issn:0372-2112.2005.03.008 [83] MARINO A, SANJUAN-FERRER M J, HAJNSEK I, et al. Ship detection with spectral analysis of synthetic aperture radar: A comparison of new and well-known algorithms[J]. Remote Sensing, 2015, 7(5): 5416–5439. doi: 10.3390/rs70505416 [84] 种劲松, 朱敏慧. SAR图像舰船及其尾迹检测研究综述[J]. 电子学报, 2003, 31(9): 1356–1360. doi: 10.3321/j.issn:0372-2112.2003.09.020CHONG Jinsong and ZHU Minhui. Survey of the study on ship and wake detection in SAR imagery[J]. Acta Electronica Sinica, 2003, 31(9): 1356–1360. doi: 10.3321/j.issn:0372-2112.2003.09.020 [85] SCHULER D L, LEE J S, HOPPEL K, et al. Polarimetric sar image signatures of gulf-stream features and ship wakes[C]. 1992 International Geoscience and Remote Sensing Symposium, Houston, USA, 1992. [86] POTTIER E, BOERNER W M, and SCHULER D L. Polarimetric detection and estimation of ship wakes[C]. IEEE 1999 International Geoscience and Remote Sensing Symposium, Hamburg, Germany, 1999. [87] WU Peng, WANG Jun, WANG Wenguang, et al. Polarimetric characters extraction research of Kelvin wakes on PolSAR image[C]. 2011 International Conference on Remote Sensing, Environment and Transportation Engineering, Nanjing, China, 2011. [88] HENNINGS I, ROMEISER R, ALPERS W, et al. Radar imaging of Kelvin arms of ship wakes[J]. International Journal of Remote Sensing, 1999, 20(13): 2519–2543. doi: 10.1080/014311699211912 [89] 叶文隽. SAR图像舰船尾迹检测研究[D]. [硕士论文], 国防科学技术大学, 2009.YE Wenjun. Research on detection of ship wake from SAR imagery[D]. [Master dissertation], National University of Defense Technology, 2009. [90] IMBO P, SOUYRIS J C, and YEREMY M. Wake detection in polarimetric SAR images[C]. IEEE 2001 International Geoscience and Remote Sensing Symposium Scanning the Present and Resolving the Future, Sydney, Australia, 2001. [91] KASILINGAM D, SCHULER D, LEE J S, et al. Modulation of polarimetric coherence by ocean features[C]. 2002 IEEE International Geoscience and Remote Sensing Symposium, Toronto, Canada, 2002: 432–434. [92] MORRIS J, ANDERSON S, and PARFITT A. Polarimetric mapping of ship wakes[C]. 2002 IEEE International Geoscience and Remote Sensing Symposium, Toronto, Canada, 2002. [93] MORRIS J and ANDERSON S. An entropy-based approach to wake echo analysis [ship wake radar detection][C]. 2003 International Conference on Radar, Adelaide, Australia, 2003. [94] YANG Jingsong, HUANG Weigen, XIAO Qingmei, et al. Optimal polarization for the observation of ocean features with SAR[C]. 2004 IEEE International Geoscience and Remote Sensing Symposium, Anchorage, USA, 2004. [95] ARNOLD-BOS A, MARTIN A, and KHENCHAF A. A versatile bistatic & polarimetric marine radar simulator[C]. 2006 IEEE Conference on Radar, Verona, USA, 2006. [96] XU Zhou, TANG Bo, and CHENG Shuiying. Faint ship wake detection in PolSAR images[J]. IEEE Geoscience and Remote Sensing Letters, 2018, 15(7): 1055–1059. doi: 10.1109/LGRS.2018.2823007 [97] BIONDI F. Low-rank plus sparse decomposition and localized radon transform for ship-wake detection in synthetic aperture radar images[J]. IEEE Geoscience and Remote Sensing Letters, 2018, 15(1): 117–121. doi: 10.1109/LGRS.2017.2777264 [98] BIONDI F. A polarimetric extension of low-rank plus sparse decomposition and radon transform for ship wake detection in synthetic aperture radar images[J]. IEEE Geoscience and Remote Sensing Letters, 2019, 16(1): 75–79. doi: 10.1109/LGRS.2018.2868365 [99] 徐丰, 王海鹏, 金亚秋, 等. 合成孔径雷达图像智能解译[M]. 北京: 科学出版社, 2020.XU Feng, WANG Haipeng, JIN Yaqiu, et al. Synthetic Aperture Radar Image Intelligent Interpretation[M]. Beijing: Science Press, 2020. [100] SAIN S R. The nature of statistical learning theory[J]. Technometrics, 1996, 38(4): 409. doi: 10.1080/00401706.1996.10484565 [101] 杜兰, 王兆成, 王燕, 等. 复杂场景下单通道SAR目标检测及鉴别研究进展综述[J]. 雷达学报, 2020, 9(1): 34–54. doi: 10.12000/JR19104DU Lan, WANG Zhaocheng, WANG Yan, et al. Survey of research progress on target detection and discrimination of single-channel SAR images for complex scenes[J]. Journal of Radars, 2020, 9(1): 34–54. doi: 10.12000/JR19104 [102] GAO Gui. A parzen-window-kernel-based CFAR algorithm for ship detection in SAR images[J]. IEEE Geoscience and Remote Sensing Letters, 2011, 8(3): 557–561. doi: 10.1109/LGRS.2010.2090492 [103] 张晓玲, 张天文, 师君, 等. 基于深度分离卷积神经网络的高速高精度SAR舰船检测[J]. 雷达学报, 2019, 8(6): 841–851. doi: 10.12000/JR19111ZHANG Xiaoling, ZHANG Tianwen, SHI Jun, et al. High-speed and high-accurate SAR ship detection based on a depthwise separable convolution neural network[J]. Journal of Radars, 2019, 8(6): 841–851. doi: 10.12000/JR19111 [104] CHEN Siwei, TAO Chensong, WANG Xuesong, et al. Polarimetric SAR targets detection and classification with deep convolutional neural network[C]. 2018 Progress in Electromagnetics Research Symposium, Toyama, Japan, 2018. [105] ZHOU Feng, FAN Weiwei, SHENG Qiangqiang, et al. Ship detection based on deep convolutional neural networks for polsar images[C]. 2018 IEEE International Geoscience and Remote Sensing Symposium, Valencia, Spain, 2018. [106] 鲁兵兵. 基于深度学习的PolSAR图像分类与舰船检测方法[D]. [硕士论文], 西安电子科技大学, 2019.LU Bingbing. PolSAR image classification and ship detection method based on deep learning[D]. [Master dissertation], Xidian University, 2019. [107] JIN Kan, CHEN Yilun, XU Bin, et al. A patch-to-pixel convolutional neural network for small ship detection with PolSAR images[J]. IEEE Transactions on Geoscience and Remote Sensing, 2020, 58(9): 6623–6638. doi: 10.1109/TGRS.2020.2978268 [108] COZZOLINO D, DI MARTINO G, POGGI G, et al. A fully convolutional neural network for low-complexity single-stage ship detection in sentinel-1 SAR images[C]. 2017 IEEE International Geoscience and Remote Sensing Symposium, Fort Worth, USA, 2017: 886889. [109] 李其. 基于深度特征的SAR图像舰船目标检测方法研究[D]. [硕士论文], 电子科技大学, 2020.LI Qi. Research of ship detection in SAR images based on depth features[D]. [Master dissertation], University of Electronic Science and Technology of China, 2020. [110] FITCH J P, LEHMAN S K, DOWLA F U, et al. Ship wake-detection procedure using conjugate gradient trained artificial neural networks[J]. IEEE Transactions on Geoscience and Remote Sensing, 1991, 29(5): 718–726. doi: 10.1109/36.83986 [111] KANG K M and KIM D J. Ship velocity estimation from ship wakes detected using convolutional neural networks[J]. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2019, 12(11): 4379–4388. doi: 10.1109/JSTARS.2019.2949006 期刊类型引用(1)
1. 毛德庆,杨建宇,杨明杰,张永超,张寅,黄钰林. IAA-Net:一种实孔径扫描雷达迭代自适应角超分辨成像方法. 雷达学报. 2024(05): 1073-1091 . 本站查看
其他类型引用(0)
-