SAR Sensor Trajectory Deviations Fourier Domain Formulation and Extended Scene Simulation

合集下载

星载-机载混合双站SAR回波信号的仿真

星载-机载混合双站SAR回波信号的仿真杨永红;凌霖【摘要】对于星载-机载双站合成孔径雷达(SA-BiSAR)系统,回波模拟的关键是计算出卫星、飞机和目标这三者之间的距离.在考虑地球自转的情况下,提出了用坐标变换的方法,解决了近地空间与外太空之间距离的计算问题,实现了任意几何模型下更为精确的SA-BiSAR回波模拟.同时,利用此方法也可获得直达波信号.点目标的仿真展示了SAR回波信号的特征,该方法得到了验证.【期刊名称】《舰船电子对抗》【年(卷),期】2010(033)006【总页数】3页(P54-56)【关键词】双站合成孔径雷达;回波模拟;成像算法【作者】杨永红;凌霖【作者单位】江苏科技大学,镇江,212003;江苏科技大学,镇江,212003【正文语种】中文【中图分类】TN9580 引言与单站SAR相比,双站SAR可以获取目标的丰富信息,提高系统的抗摧毁、抗隐身性能等优点。

双站SAR的载体可以是飞机、卫星以及固定的高塔。

星载-机载混合BiSAR（SA-BiSAR）是由卫星作为发射平台、飞机作为接收平台所组成的。

在没有实测数据的情况下,为了对SA-BiSAR系统进行深入的研究,利用仿真工具来建立回波信号是一个必要且可行的途径。

回波信号模拟是SAR的一个专门的研究方向。

Franceschetti研究了分布式场景在频域内SAR原始信号的仿真问题,在原始信号中包含了SAR载体的轨迹偏差,同时给出了限制条件,该方法得到了验证。

本文在考虑地球自转的情况下,提出了用坐标变换的方法,解决了近地空间与外太空之间距离的计算问题,并以点目标的仿真为例验证了该方法的可行性。

1 几何模型对于SA-BiSAR的几何模型,假设地球为一个均匀的球体,即地表高度为零；卫星的轨迹用轨道坐标系来描述；地球上的目标具有一定的速度,这来自于地球自转的影响。

同时,飞机在近地空间中飞行,也将受到地球自转的影响。

回波模拟的关键是计算出卫星、飞机和目标这三者之间的距离（卫星-飞机、飞机-目标、卫星-目标）。

基于Fourier原子的基追踪方法在SAR超分辨成像中的应用

ｐｒｕｔＣｅｉｌｍｅｔｄｓｅｙａｄｓａｌ．ａｄｉＣｒｖｄｕｒｒｓｌｔｎａｈｏＵｉ．ｕｓｉａｂｎｍｐｅｎｅｐｄｎｔｙｎｔａｐｉｅｓｐ－ｅｏｕｉｔｅｓＪｅｔｅｂｎｏｅｏｔｍｅ
维普资讯
第２６卷第３期
２００７年６月文章编号：０１９１（０７００９０１０ — ０４２０）３— １６— ５
红外与毫米波学报
ＪｎｒｒｄＭｉｉ．Ｗａｅ．Ｉｆａｅｌｍｌｖｓ
２７６０Ａｍ，Ｇｉｎ５１０，ｈｎ）．５６ｒｙｕｌ４０２Ｃｉａｉ
ＡｂｔａｔｈｐｌａｉｎｏｅｂｓｓｐｒｕｔｍｅｈｄｔＡＲｓｐｒｒｓｌｔｎｐｏｅｓｇｗａｘｅｄｄｓｒｃ：ＴｅａｐｉｔｆｈａｉｕｓｉｃｏｔｔｏＳｕｅ－ｅｏｕｉｒｃｓｉｓｅｔｎｅ．Ｆｒｔｏｏｎｉｌｓｙ，ｂｅｎｓａｄｏｈＡＲｔｉｕｅｃｔｒｎｄｌｏｒｒｉｔｎｒｉａｔｍｐｉｉａｇｒｍａｅｉｎｄｉｈｅｈｓｏｙｄｍａｎｔｅＳａｔｂｔｄｓａｔｉｇｍｏｅ，ａＦｕｅｃｉａｙｗｔｆｓｌｔｌｏｔｗｓｄｓｇｅｐａｉｒｏｉ．ｒｅｉｄｏｈｉｃｉｈｎｓｔ
ｒｈａｒｏｅｏｖｅｏｔｉｔｎｐｏｌｍｏｅｂｉｐｒｕ．Ｆｎｌ，ｉｅｓｌｔｎｉｇａｂａｅｙｉｍｗｓｏｓｄｔｓｌｔｐｉｚｉｒｂｅｆａｓｕｉｉａｙｈｒｅｏｕｏｅＷｏｔｎｄｂｔｐｐｏｅｈｍａｏｈｔｓｓｔｌｇｒｈｉｍａｓｉ

(完整版)各种SAR成像算法总结,推荐文档

sr
(t)
Wa
t
R(t) c
st
t
2R(t) c
(1.20)
n
Wa
t
R(t) c
p
t
nPRT
2R(t) c
其中，为目标的后向散射特性，Wa (A) 为方位向的天线方向性函数， c 为光速。
sr (t) 经正交解调后的复信号 s(t) 可以表示为：
s(t)
n
Wa
t
R(t) c
s0
t
1.2 SAR 回波信号模型
1.1 节分析了 SAR 成像的基本原理，本节推导 SAR 回波信号的数学模型，
给出 SAR 信号处理的理论基础。
chirp 信号是 SAR 系统中最常用的发射信号形式。假设雷达发射的 chirp 脉
冲串 st (t) 为：
n
st (t) p(t nPRT ) n
(1.19)
1.1 SAR 成像原理
本节以基本的正侧视条带工作模式为例，对 SAR 的成像原理进行分析和讨
论。
正侧视条带 SAR 的空间几何关系如下图所示。图中，αoβ 平面为地平面，
oγ 垂直于 αoβ 平面。SAR 运动平台位于 S 点，其在地面的投影为 G 点。SAR
运动平台的运动方向 Sx 平行于 oβ，速度大小为 va 。SAR 天线波束中心与地面的交点为 C，CG 与运动方向 Sx 垂直；S 与 C 的距离为 Rs ， B1SB2 称为天线波束的方位向宽度，大小为 a 。P 为测绘带内的某一点，一般情况下取斜距平面 CSP 进行分析，称 SAR 运动的方向 Sx 为方位向（或方位维），称天线波束指向
量 fd (t) 为：
fd

一种利用邻域相对熵的SAR影像变化检测方法

第33卷第3期 2018年6月遥感信息Remote Sensing InformationVol, 33，N o.3Jun. ,2018一种利用邻域相对熵的SAR影像变化检测方法刘本强1>2,赵争2，魏钜杰2(1.山东科技大学测绘科学与工程学院，山东青岛266590;2.中国测绘科学研究院，北京100830)摘要：为了解决传统代数计算法构造的差异图背景中含有较多噪点的问题，提高变化检测的精度，弓丨入信息论中相对熵的概念，借助邻域处理，提出了一种基于邻域相对熵的差异图构造方法，并应用模糊局部信息C均值 (fuzzy local information C-means，FLICM)非监督聚类算法，实现变化信息的自动提取。

通过采用4组单极化前后时相SAR影像数据集，分析对比了不同邻域形式的相对熵差异图和传统差异图的检测性能。

实验结果表明，应用该方法生成的差异影像，对噪声有着较强的鲁棒性，能够满足变化检测的需求，且在定量评价的性能指标方面表现较好。

其中，基于D■邻域相对熵差异图进行变化检测的结果更加突出。

关键词：合成孔径雷达；邻域；相对熵；模糊局部信息C均值；变化检测doi：10. 3969/j.issn.1000-3177. 2018. 03. 014中图分类号：TP75 文献标志码：A文章编号=1000-3177(2018)157-0091-07A Method of SAR Image Change Detection Based onRelative Entropy Within NeighborhoodLIU Benqiang1'2 ,ZH A O Zheng2 ,W EI Jujie2(1. Shandong University o f Science and Technology ,College o f Geomatics 9Q ingdao9Shandong266590,CAina；2. Chinese Academy of Surveying and Mapping，Beijing 100830，China)Abstract:The traditional change detection method has a common problem that speckle noise affects the quality of the difference image. In order to solve this problem and improve detection accuracy?a novel unsupervised change detection method based on neighborhood relative entropy and fuzzy local information C-means (FL IC M) is introduced in this paper. Using the proposed method, change detection map can be obtained automatically. Four single-polarization SAR datasets are used to verify our method. Experimental results show that the proposed method gains a better performance and is more robust with the influence of noise. Especially,using relative entropy within D-neighborhood,the difference image has a better result than other methods.Key words：S A R；neighborhood;relative entropy；fuzzy local information C-meansjchange detection〇引言获取地表变化信息，一直以来是遥感领域研究的热点问题。

曲线运动sar成像算法研究

曲线运动sar成像算法研究曲线运动SAR成像是近几年新型的一种成像技术。

它是将目标物体轨迹在空间中曲线转化成距离时间变化模式表示，并借助SAR 技术获得物体图像信息。

由于曲线运动SAR成像具有阵列排列紧凑、成像效率高、探测范围可调等特点，在无人机目标跟踪、地面目标识别等应用领域具有重要的战略意义和重要的应用价值。

一、曲线运动SAR成像原理曲线运动SAR成像是把物体运动轨迹转化为时间序列的距离变化模式，然后借助SAR技术，对时间序列的距离变化模式进行采样，获得物体图像信息。

曲线运动SAR成像由以下四个步骤组成： 1.迹参数估计：轨迹参数估计主要是用来确定物体运动轨迹及其有关参数，如最大运动速度、运动方向等。

2. matlab模拟时间序列：在估计出物体运动轨迹的参数后，利用Matlab的曲线拟合算法拟合曲线，确定物体在不同时间的位置，得到物体在空间中的时间序列，从而得出物体的距离时间变化模式。

3. SAR信号采样：根据物体的距离时间变化模式，对其采样，获得信号采样序列，从而得到物体的位置信息。

4.像恢复：将采样得到的信号序列进行预处理，然后使用图像恢复技术，将目标图像恢复出来。

二、曲线运动SAR成像算法曲线运动SAR成像算法的核心是将物体的运动轨迹转化为时间序列的距离变化模式，并借助SAR技术获得物体图像信息。

具体算法步骤如下：1.择运动轨迹模型：首先选择最适合物体行走轨迹的模型，包括线性运动模型、圆弧运动模型和组合运动模型等；2. 估计轨迹参数：利用模型估计物体的运动轨迹的参数，如最大运动速度、运动方向等；3. matlab模拟时间序列：利用Matlab的曲线拟合算法，确定物体在不同时间的位置，获得物体的时间序列，从而得出距离时间变化模式；4. SAR信号采样：根据物体的距离时间变化模式，对其采样，获得信号采样序列，并由此确定物体的位置；5.像恢复：将采样得到的信号序列进行预处理，然后使用图像恢复技术，将目标图像恢复出来。

Comparison of Topography- and Aperture-Dependent Motion Compensation Algorithms for Airborne SAR

Comparison of Topography-and Aperture-Dependent Motion Compensation Algorithmsfor Airborne SARPau Prats,Member,IEEE,Karlus A.Câmara de Macedo,Andreas Reigber,Member,IEEE,Rolf Scheiber,and Jordi J.Mallorqui,Member,IEEEAbstract—This letter presents a comparison between three Fourier-based motion compensation(MoCo)algorithms for airborne synthetic aperture radar(SAR)systems.These algo-rithms circumvent the limitations of conventional MoCo,namely the assumption of a reference height and the beam-center approx-imation.All these approaches rely on the inherent time–frequency relation in SAR systems but exploit it differently,with the con-sequent differences in accuracy and computational burden.After a brief overview of the three approaches,the performance of each algorithm is analyzed with respect to azimuthal topography accommodation,angle accommodation,and maximum frequency of track deviations with which the algorithm can cope.Also,an analysis on the computational complexity is presented.Quantita-tive results are shown using real data acquired by the Experimen-tal SAR system of the German Aerospace Center(DLR).Index Terms—Calibration,image registration,interferometry, motion compensation(MoCo),synthetic aperture radar(SAR).I.I NTRODUCTIONM OTION compensation(MoCo)represents a key issue in the focusing of airborne synthetic aperture radar (SAR)data.The fact that the platform does not follow an ideally linear trajectory turns into a severe degradation of the impulse response function(IRF)if the deviations are not considered.Hence,the SAR focusing operation must ensure a proper correction of such deviations by means of a modiﬁed processing scheme.Moreira et al.[1]proposed the two-step MoCo approach integrated in the extended chirp scaling(ECS) algorithm,which allows the retrieval of accurate interferometric products with airborne systems.This two-step MoCo consists of the application of a phase and envelope correction in the beginning of the processing(ﬁrst-order MoCo),where the deviations are corrected for a reference range and height.A range-dependent correction is not possible at this stage since range cell migration(RCM)is still present in the data.A secondManuscript received September29,2006;revised January29,2007.This work was supported in part by the Spanish MCYT and FEDER funds under Project TEC2005-06863-C02-01.The work of K.A.Câmara de Macedo is supported by a grant from CAPES,Brazil.P.Prats,K.A.Câmara de Macedo,and R.Scheiber are with the Microwaves and Radar Institute,Deutsche Zentrum für Luft-und Raumfahrt,82234 Oberpfaffenhofen,Germany(e-mail:pau.prats@dlr.de;karlus.macedo@dlr.de; rolf.scheiber@dlr.de).A.Reigber is with the Berlin University of Technology,10587Berlin, Germany(e-mail:anderl@cs.tu-berlin.de).J.J.Mallorqui is with the Universitat Politècnica de Catalunya,08034 Barcelona,Spain(e-mail:mallorqui@).Digital Object Identiﬁer10.1109/LGRS.2007.895712range-dependent phase correction is applied after range com-pression and RCM correction(RCMC),and before azimuth compression(second-order MoCo).In this step,topography variations can be partially considered using an external digital elevation model(DEM),but the correction to be applied also depends on the position of the target within the -monly,a reference height is assumed for the whole scene,while the correction is applied for the center of the beam.In[2],a detailed analysis of this latter approximation is presented. Three algorithms that circumvent these two limitations have been proposed in the literature[3]–[6].They all rely on the time–frequency relation of the azimuth SAR signal but exploit it in different ways.Also,they use an external DEM to compute the proper phase delays to be applied to each target to correctly accommodate topography variations.The goal of this letter is to analyze the performance of each algorithm based on the accuracy in the accommodation of the different parameters of interest:topography,azimuth angle,and frequency of track deviations.Furthermore,a quantitative analysis is carried out by comparison with a time-domain SAR processor,which allows applying MoCo with no approximation.Section II starts with a review of the limitations in con-ventional MoCo when applied inside Fourier-based processors. Section III revisits the three algorithms under study by pre-senting their main characteristics,which will allow a proper analysis and comparison of the results.Section IV makes a statement regarding the phase content of an interferometric pair when processing using the presented approaches.Section V shows the results with real data acquired by the Experimental SAR(E-SAR)system of the German Aerospace Center(DLR). Finally,Section VI brings the conclusions.II.L IMITATIONS IN C ONVENTIONAL M O C OIt is well known that the conventional two-step MoCo makes two approximations[3].1)Topography approximation:A reference height for thewhole scene is assumed to apply MoCo,which yields IRF degradation for targets at a height different than the refer-ence one.An external DEM can be used in second-order MoCo,but this approach cannot accurately accommodate topography if it varies signiﬁcantly along the synthetic aperture,which,depending on resolution and wavelength, can be as large as several hundred meters.1545-598X/$25.00©2007IEEE2)Beam-center approximation:The correction also dependson the relative position of the target with respect to the sensor,i.e.,the correction depends on the azimuthal aperture/ually,the correction is applied at the center of the beam,which results in IRF degradation when the processed azimuthal beamwidth is large.This degradation has more impact in low frequency systems, where a wide aperture is necessary to achieve a high azimuthal resolution[7].The effects of both approximations are similar and well established in the literature[2],[3].These are mainly a shift of the IRF,defocusing,and phase aberrations.These will turn into signiﬁcant phase and coregistration errors in the generated interferogram.Note that in interferometric single-pass sys-tems,the deviations are correlated in master and slave images; hence,the errors tend to cancel out after interferogram genera-tion.However,interferometric repeat-pass products are usually affected by these approximations.Normally,theﬁrst of the commented approximations is the one that has more impact,but for high-precision SAR focusing,both approximations must be considered.III.T OPOGRAPHY AND A PERTURE C OMPENSATIONTo apply a topography-and aperture-dependent MoCo,the algorithms rely on the well-known time–frequency relation of the azimuth SAR signal.The mapping between time and frequency is expressed byf a(t;r0,t0)=2vλsinβ=2v2λt0−tr20+v2·(t−t0)2(1)where t is the azimuth time,βis the azimuth angle correspond-ing to an azimuth frequency f a,v is the forward velocity of the platform,t0is the zero-Doppler time,r0is the closest approach distance,which corresponds to t0,andλis the used wavelength. The correction to be applied is both azimuth time dependent and azimuth angle/frequency dependent.An efﬁcient solution to this problem can be achieved via block processing,which becomes the tool used by the three approaches presented next.A.Precise Topography-and Aperture-Dependent(PTA)AlgorithmThe PTA algorithm was presented in[3].It is applied with the image already focused so that the residual phase correction is computed for each target considering its true topographic height and the real distance at each time instant,hence also accommodating the azimuth angle variation.The correction is applied in the frequency domain using the time–frequency mapping indicated by(1),in principle just by reversing the cor-rection vector computed in the time domain.The implemented approach makes only one forward Fourier transform(FT)per block,taking a50%overlap(α=0.5)to avoid discontinuities at block borders.An inverse FT is needed for each target after the correction,which results in a relatively high computational burden.(Since only one sample after the inverse FT is kept,the direct computation of that single value can save some time.)ForTABLE IP ROPERTIES OF THE A LGORITHMS U NDER S TUDY W ITH R ESPECT TO TOPOGRAPHY,A NGLE,AND R ESIDUAL E RRORS A CCOMMODATIONefﬁciency purposes,the block length should be much smaller than the synthetic aperture,e.g.,32or64azimuth samples. Also,higher efﬁciency can be gained by changing adaptively the block size depending on track and topography variations. The main advantage of this algorithm lies in the fact that it can accommodate the scene topography within the sample spacing.On the other hand,given that the correction is down-sampled to the length of the block,the ability to accommodate both the angle and the high-frequency component of the resid-ual error is reduced.These drawbacks can be circumvented by increasing the block size N.However,this increases the computational burden,hence showing a clear tradeoff between accuracy and efﬁciency.B.Subaperture Topography-and Aperture-Dependent (SATA)AlgorithmThe SATA algorithm wasﬁrst presented in[4].Again,accu-rate MoCo can be obtained by exploiting the time–frequency relation in SAR.The main difference with respect to PTA is that in this case the correction is applied before azimuth com-pression.The solution proposed in[4]is to use the short-time FT in blocks(or subapertures)along the azimuthal direction. The block size N is a compromise between topography/angle accommodation and trajectory deviation accommodation.A larger block size increases the accuracy of the topography/angle accommodation.However,since the correction to be applied considers the track deviation in the middle of the block,the accommodation of fast track variations is reduced in this case.A small block size performs in the opposite way.A clear advantage of this algorithm is that it is computationally very efﬁcient;the number of necessary FTs is drastically reduced when compared to the other two algorithms.It should be noted that the accommodation of the topography depends on range(see Table I);hence,a better resolution is achieved at near range than at far range.A straightforward so-lution to this problem is to increase the block length with range. Similar to PTA,adaptive approaches for selecting the block size can be implemented.Also,some overlap between blocks should be considered to reduce the appearance of sidelobes due to phase jumps in the phase history of targets.C.Frequency Division(FD)AlgorithmAn algorithm based on multiple look processing wasﬁrst presented in[5]to accommodate aperture-dependent motion errors,while in[6]a very similar algorithm(i.e.,FD)was suggested,which also makes a comment on topography accom-modation.It takes a similar approach as the SATA algorithm but makes the blocks(looks)in the frequency domain instead of in the time domain.Therefore,just before azimuth compressionPRATS et al.:COMPARISON OF TOPOGRAPHY-AND APERTURE-DEPENDENT MoCo ALGORITHMS 351TABLE IIC OMPUTATIONAL B URDEN OF THE A LGORITHMS U NDERS TUDY W ITH R ESPECT TO FFT OPERATIONSand after conventional second-order MoCo,data are Fourier transformed in the azimuthal direction.Afterward,only a small block in frequency (a look)with size N is selected each time,which needs to be zero padded up to the original size of the data N az .The central frequency of this look corresponds to a certain squint,which is used to compute the appropriate correction.Therefore,after an inverse FT,a phase correction can be applied considering each time instant of the track deviations.Finally,a coherent addition of the different looks can be performed before the ﬁnal azimuth compression step.The performance of the FD is as follows.On one hand,the length of the look N constrains the accommodation of the angle along the beam.Since the correction is applied considering the middle frequency in the look,a longer block will result in a worse angle accommodation.On the other hand,N also deﬁnes the resolution in the accommodation of the topography,since the bandwidth of the look results in a certain synthetic aperture sublength.Hence,many small looks are preferred in the FD algorithm.The main advantage of FD is that it can accom-modate track deviations accurately as the correction is applied at each time instant,although at the expense of computational burden.For every look,an inverse FT of N az points has to be performed,and the MoCo phase correction has to be applied for every look to the whole data mentsEach algorithm can be properly characterized using the for-mulas shown in Table I,namely the resolution in the accom-modation of both the topography along azimuth δx and the aperture angle δθaz ,and the maximum frequency of the residualerror f resmaxwith which the algorithm can cope after ﬁrst-and second-order MoCo.These formulas have been obtained using basic SAR and signal processing relations.PRF is the effective pulse repetition frequency used during processing,θaz is the processed aperture beamwidth,N is the number of azimuth samples per block,L sa is the length of the synthetic aperture in seconds,r is the closest approach range distance of the target,α(0≤α<1)is an overlap factor,and N az is the number of azimuth samples of the data matrix.Table II shows the computational burden considering only the operations related to the fast FT (FFT)for one azimuth line.The formulas shown in Tables I and II are an excellent tool to analyze the performance of each algorithm.Fig.1plots these formulas for different values of the block size N .The system parameters in Table III have been used in this case,assuming a target at mid range.It is clear that PTA gives the best accommodation of topography,while FDcanFig.1.Plots of the properties of the algorithms for different processing block sizes.(a)Topography accommodation accuracy δx .(b)Aperture angle accommodation accuracy δθaz .(c)Maximum frequency of residual errors f res max.(d)Computational burden for an azimuth line.TABLE IIIM AIN S YSTEM AND P ROCESSING PARAMETERSbetter accommodate fast track deviations.The performance of SATA seems to lie between these two,with a clear advantage concerning the computational burden.It is interesting to note that PTA is the only one that can achieve the best performance when the size of the block is equal to that of the synthetic aperture in samples (N =L sa ·PRF ).Obviously,this will yield an important increase in computational burden.Nevertheless,all these algorithms focus on the azimuth signal,assuming RCMC has been properly carried out.It should be noted that the requirements for fast track de-viations are normally not very stringent,since the frequency of the deviations experienced by current airborne platforms is well under the limits shown in Table I,e.g.,the maximum fre-quency of track deviations for the E-SAR system is ∼4Hz [8].Furthermore,the fact that ﬁrst-and second-order MoCos have already been applied reduces even more the frequency of the needed correction.A fast variation where FD might perform better than the other two algorithms,e.g.,several cycles of high amplitude within a synthetic aperture,is not realistic for current airborne platforms.IV .C ONSIDERATIONS IN THE I NTERFEROMETRIC C ASE It is of vital importance to understand the interferometric phase content of an interferometric pair after applying any352IEEE GEOSCIENCE AND REMOTE SENSING LETTERS,VOL.4,NO.3,JULY2007 of the approaches described above.In[9],it is shown thattheﬂattened phase after applying MoCo assuming the sameconstant reference height for the whole scene isφﬂat=−4πλB real cos(θ−αreal)r t0sinθh=k realz·h(2)whereθis the off-nadir look angle,r t0is the distance from the real antenna position,B real andαreal are the real baseline and tilt angle corresponding to the real trajectory,and h is the height of the target with respect to the reference height assumed during MoCo.Equation(2)also assumes a constant deviation from the reference track along the synthetic aperture so that the values refer to the beam-center position.Indeed,this is also the case if a DEM is used during MoCo. Assuming a perfect MoCo,the external DEM should be sub-tracted from the interferometric phase just in the same way as theﬂat earth is subtracted.Doing so,the residual phase is proportional to the DEM error h error and sensitive to the real baselineφres(x,r)=k realz(x,r)·h error(x,r).(3) With this information,it is straightforward to correct the origi-nal DEM.An important consideration in the case of using SATA or FD algorithms must be stated.Given that the topography accommodation accuracy is limited by the block size,the external DEM should be smoothed accordingly before using it to avoid wrong height estimates.This step will be necessary whenever the DEM resolution is better than that of the algo-rithms,where the latter is given by the values ofδx in Table I.V.R ESULTSTo compare the three approaches,L-band data acquired by the E-SAR system of DLR in a repeat-pass mode are used.The data have been processed with both ECS[1]and a time-domain backprojection algorithm.The latter is used for accuracy assess-ment as it applies topography-and aperture-dependent MoCo with no approximation.Hence,the accuracy of each algorithm can be estimated for a given scenario after comparison with the backprojection output.The data have beenﬁrst processed using a Shuttle Radar Topography Mission(SRTM)DEM back geocoded to slant-range geometry.The spatial resolution of this DEM(90×90m)is low compared to the airborne image resolution.It is recommended to use efﬁcient cubic convolution algorithms to obtain a smoothed interpolated DEM.Otherwise, SATA and FD will result in phase artifacts close to slope discon-tinuities[4].The data have also been processed with a5×5m resolution DEM obtained by the E-SAR at X-band in single-pass mode.Therefore,the accuracy in the accommodation of topography variations can be tested with both low-and high-resolution DEMs.Table III summarizes the main system and processing parameters.The maximum horizontal and vertical deviations of the platform are±6and±5m,respectively. Fig.2shows the reﬂectivity image and the SRTM DEM of the observed scene,where it can be noted that the scene has topographic variations of about1000m.The output foreach Fig.2.(a)Reﬂectivity image and(b)back geocoded DEM in meters over the WGS-84ellipsoid of the observed scene.Scene dimensions:7×2km (azimuth×range).Near range on top of theimage.Fig.3.Phase error(N=32)for(a)PTA,(b)SATA,and(c)FD.of the topography-and aperture-dependent algorithms has been compared with the output of the time-domain backprojection algorithm by means of a simple Hermitian product.Fig.3 shows the phase error for each algorithm when using the SRTM DEM.It is clear that PTA has the best performance,as high topographic variations are accurately accommodated.On the other hand,both SATA and FD have problems to accommodate areas with high topographic changes;for example,in some places,the elevation changes are larger than300m in lessPRATS et al.:COMPARISON OF TOPOGRAPHY-AND APERTURE-DEPENDENT MoCo ALGORITHMS 353TABLE IVC OMPUTATIONAL B URDEN AND ACCURACYthan 40azimuth samples.However,in general,the performance of all three algorithms can be considered excellent.Table IV shows the computational burden to process one image with each of the algorithms for different block sizes.The phase error stan-dard deviation when compared to the time-domain processor is also shown for both SRTM and X-band DEMs.The result with conventional MoCo,i.e.,using the beam-center approximation and using the external DEM during second-order MoCo,is also included.Concerning the high-resolution X-band DEM,it is clear again that PTA shows the best performance,while SATA and FD fail to efﬁciently use the external information of the DEM.Note that the computational burden decreases in some cases when the block size increases,which contradicts the formulas given in Table II.This is because Table II only considers the computational burden due to FFTs.However,other operations are carried out inside each block,e.g.,true distance computa-tion,complex products,etc.,which have an important inﬂuence on the ﬁnal processing time.With regard to SATA,the error increases dramatically with the block size.As already commented,this is because the cor-rection is applied considering the track position in the middle of the block.With a larger block,a bigger error is introduced at the edges,degrading in this way the result.It is worth mentioning that the accuracy of the PTA does not vary signiﬁcantly with an increasing block size,as shown in Table IV.This suggests that the residual errors to correct have low frequency components in this data set,so that even with a block size of 32samples,they can be accommodated with PTA.Hence,increasing the block size does not perceptively improve the result in this case.VI.C ONCLUSIONThis letter has shown the comparison of different algorithms to accommodate topography-and aperture-dependent motion errors in airborne systems.These methods are thought to be implemented together with Fourier-based processors,such as ECS [1],range Doppler,or ω−k [10].The need of such MoCo algorithms becomes evident in repeat-pass systems,where track deviations are not correlated,degrading in this way the retrieved interferogram.After the theoretical and experimental results,it is possible to obtain clear conclusions.Concerning accuracy,PTA is the clos-est one to the ideal case.Indeed,with a block size equal to the synthetic aperture,it can accurately correct all phase errors.On the other hand,SATA shows the best accuracy–computational burden ratio.With a negligible computational cost,it achieves reasonable accuracy,which is acceptable in most scenarios with low to moderate topography.Note that SATA can result in large errors in airborne systems where fast track deviations can occur.In such cases,it is recommended to use PTA with a larger block size,or alternatively FD.Finally,FD results in a high computational burden when high accuracy is desired.The main advantage of this approach is that it can perfectly accommodate fast track variations.However,the computational effort is higher than for PTA for a similar topography accom-modation resolution.Although only one data set is shown in this letter,similar performances have been obtained in other scenarios.It has also been shown that the residual phase of an inter-ferometric pair is proportional to the DEM error and the real baseline.Therefore,a smoothing of the DEM should be applied with SATA or FD whenever the DEM azimuthal resolution is better than that achievable by these algorithms.Future work includes the optimization of the processing chain to reduce the errors that can be observed in the PTA case.Additionally,the comparison with fast time-domain processors,such as [11],will be a point to address in the future.A CKNOWLEDGMENTThe authors would like to thank the anonymous reviewers for their valuable comments and suggestions.R EFERENCES[1]A.Moreira,J.Mittermayer,and R.Scheiber,“Extended chirp scalingalgorithm for air-and spaceborne SAR data processing in stripmap and scanSAR imaging modes,”IEEE Trans.Geosci.Remote Sens.,vol.34,no.5,pp.1123–1136,Sep.1996.[2]G.Fornaro,G.Franceschetti,and S.Perna,“On center-beam approxima-tion in SAR motion compensation,”IEEE Trans.Geosci.Remote Sens.,vol.3,no.2,pp.276–280,Apr.2006.[3]K.A.C.de Macedo and R.Scheiber,“Precise topography-and aperture-dependent motion compensation for airborne SAR,”IEEE Trans.Geosci.Remote Sens.,vol.2,no.2,pp.172–176,Apr.2005.[4]P.Prats,A.Reigber,and J.J.Mallorqui,“Topography-dependent motioncompensation for repeat-pass interferometric SAR systems,”IEEE Trans.Geosci.Remote Sens.,vol.2,no.2,pp.206–210,Apr.2005.[5]R.Scheiber and V .Bothale,“Interferometric multi-look techniques forSAR data,”in Proc.IEEE IGARSS ,Toronto,ON,Canada,Jun.24–28,2002,vol.1,pp.173–175.[6]X.Zheng,W.Yu,and Z.Li,“A novel algorithm for wide beam SARmotion compensation based on frequency division,”in Proc.IGARSS ,Denver,CO,Jul.31–Aug.42006,pp.3160–3163.[7]S.N.Madsen,“Motion compensation for ultra wide band SAR,”in Proc.IGARSS ,Sydney,Australia,Jul.9–13,2001,vol.3,pp.1436–1438.[8]S.Buckreuss,“Motion errors in an airborne synthetic aperture radar sys-tem,”Eur.Trans.Telecommun.Relat.Technol.,vol.2,no.6,pp.55–64,Nov.1991.[9]D.R.Stevens,I.G.Cumming,and A.L.Gray,“Options for airborneinterferometric SAR motion compensation,”IEEE Trans.Geosci.Remote Sens.,vol.33,no.2,pp.409–420,Mar.1995.[10]A.Reigber,E.Alivizatos,A.Potsis,and A.Moreira,“Extended wave-number domain SAR focusing with integrated motion compensation,”Proc.Inst.Electr.Eng.—Radar Sonar Navig.,vol.153,no.3,pp.301–310,Jun.2006.[11]L.M.H.Ulander,H.Hellsten,and G.Stenstr,“Synthetic aperture radarprocessing using fast factorized back-projection,”IEEE Trans.Aerosp.Electron.Syst.,vol.39,no.3,pp.760–776,Jul.2003.。

压缩感知SAR成像中的运动补偿

压缩感知SAR成像中的运动补偿徐建平;皮亦鸣【期刊名称】《电子与信息学报》【年(卷),期】2012(034)002【摘要】由于其具有压缩采样特性,压缩感知在高分辨SAR成像技术中得到了广泛应用.然而作为一种参数化的成像方法,基于压缩感知的成像方法对位置误差非常敏感.位置误差会造成图像偏离真实位置、散焦、甚至根本不能成像.该文针对SAR压缩成像系统中存在的运动误差,分析了平台非理想运动对回波信号的调制机理和运动相位误差对信号稀疏表征的影响,提出了基于传感器测量数据进行运动补偿的压缩感知SAR成像方法,通过在稀疏矩阵中引入附加项完成空不变运动误差的补偿.该方法不仅能以少量的测量孔径和测量数据获得重建目标空间的足够信息而且能有效降低运动误差对成像质量的影响,实现高分辨成像.%Because of its compressed sampling property, Compressive Sensing (CS) has wide application to high resolution imaging. But as a parametric imaging method, CS based imaging methods are sensitive to position error. Position error may cause defocusing, range migration, or even can not imaging. This paper concerns the analysis and compensation of trajectory deviations in airborne spotlight Synthetic Aperture Radar (SAR). A motion compensated CS SAR imaging scheme is proposed based on the sensor measured data. An additional item is introduced into the sparse matrix to achieve the compensation of space invariant motion error. This method can not only get enough information for imaging using few measure position and data,but also reduce the affection of motion error on image quality, achieve high resolution imaging.【总页数】6页(P294-299)【作者】徐建平;皮亦鸣【作者单位】电子科技大学电子工程学院成都 611731;电子科技大学电子工程学院成都 611731【正文语种】中文【中图分类】TN957.52【相关文献】1.运动补偿与UWB SAR中RD成像算法结合的实现方法 [J], 郭微光;刘光平;董臻;梁甸农2.一种ISAR二维压缩感知成像的运动补偿方法 [J], 俞翔;朱岱寅3.压缩感知在城区高分辨率SAR层析成像中的应用 [J], 廖明生;魏恋欢;汪紫芸;Timo Balz;张路4.机载SAR成像运动补偿中陀螺仪量化误差影响 [J], 孙国雯;梁兴东5.基于运动补偿和正交解耦合的双基SAR成像算法 [J], 李梦慧;谭鸽伟;杨晶晶;徐熙毅;李彪因版权原因，仅展示原文概要，查看原文内容请购买。

第五章 SAR干涉测量(完整)第五章 SAR干涉测量(完整)中科院微波遥感

SAR干涉测量Ground range Slant range AzimuthA nt e nn af o ot p ri n tS t r i p-ma pS a te ll it e o rb it1S a te ll it e o rb it2InterferometerBaselinePerpendicular BaselineInterferometric system: bistatic (SRTM 2000)y (ground range)zS 1rθHS 2P)exp()(0t j t u ω−c P r P r t n P j t j t u P s s /))()(()())(exp())(exp()()(11111011rr rr +=+−−−∝τϕτωτcP r P r t n P j t j t u P s s /))()(()())(exp())(exp()()(21222022rr rr +=+−−−∝τϕτωτThe SAR interferogram is generated by:multiplying pixel by pixel the first SAR image times the second one complex conjugated.Thus, the interferogram amplitude is the amplitude of the first image times that of the second one.Whereas its phase (called interferometric phase) is the phase difference between the two images.由于入射角的差异使得两幅图像不是完全重合，对它们进行配准处理后，配SAR interferometric phase: geometriccontributionSAR interferometric phase: geometric contribution AzimuthS l a n t r a n g eB r 1r 2Az i m u t h G r ou n d r a n g eB nrPhase ∆≈λπ4Mt. Vesuvius, baseline 250 m.Mt. Vesuvius, baseline 50 m.S l a n t r a n g eAzimuthThe role of the normal baseline The role of the normal baseline S l a n t r a n g eAzimuthBn =250The normal (or perpendicular) baseline is a keyparameter in SAR interferometryThe normal (or perpendicular) baseline is a key parameter in SAR interferometry Bn =50⊥−=∆B R z 02sin 2θλπ辅图像SLC主图像SLCDEM数据选择的原则：基线距与时间相干⎪⎧≤≤10,)sin(sin r µπµ距离向配方位滤波前后方位频谱对比（a）主像滤波前；（b）辅图像滤波前；（c）主图像滤波后；（d）辅图像滤波后A “cleaned”interferogram is achieved by averaging in areas of uniform phase. SNR improves ∝the number of looks. Usually the averaging window is adaptive.Noise sources in interferometry干涉相干All images from Space Shuttle (SIR-C) span Apr-OctFrom: Rosen et al., 1996干涉纹图干涉相干图Amplitude Phase Coherencehundred of meters). Coherence maps cannot measure this noise.Atmospheric artifacts can be up to two fringes. This is converted in elevation error, depending on the baseline. The error cannot beestimated or recovered.Atmospheric artifactsEtna Atmospheric artifacts can be detected in a multi-baseline environment, since atmosphere is incorrelated in time.Paris (+ reflectivity)利用枝切法设置的枝切线干涉纹图与利用Goldstein 算法得到的解缠结果-3-2-10123-3-2-1012301020304050-3-2-1123Original phase Irrotational phaseRotational phase Unwrapped irr. phase（a) 513×513的干涉纹图；(b)基于FFT的无权重的最小二乘算法相位解缠结果；(c)相干图；(d) Goldstein枝切法相位解缠结果实验区1:50000地形图地理编码后的数字高程模型新疆卡拉喀什地区三维透视图三维透视图±6m 90% vertical errorvertical accuracy (relative)±16m 90% vertical error vertical accuracy (absolute)±15m 90% circular error horizontal accuracy (relative)±20m 90% circular error horizontal accuracy (absolute)16-bit Signed Integer Data formatWGS84 or MSL (optionally)Datum (vertical)WGS84Datum (horizontal)1m Height levels 1"x1" Lon& Lat Raster size SRTM DEM ProductSalt Lake City, Utah差分干涉测量。

FRFT在SAR运动目标定位中的应用

FRFT在SAR运动目标定位中的应用
易航;范崇祎n;黄晓涛
【期刊名称】《雷达科学与技术》
【年(卷),期】2010(8)4
【摘要】传统SAR成像技术产生的运动目标图像是散焦的,且其成像位置偏离真实位置,为了得到运动目标的真实运动及位置信息,必须对其重新定位.运动目标的定位关键在于多普勒参数的估计,分数阶Fourier变换作为一种时频分析方法已经被广泛用于运动目标的多普勒参数估计.对分数阶Fourier变换应用于运动目标多普勒参数的估计中的特点进行了详细的分析与推导,并将其与能量均衡法结合,得到了一种不需要成像即可定位运动目标的算法,并通过仿真实验验证了该算法的有效性.【总页数】6页(P329-334)
【作者】易航;范崇祎n;黄晓涛
【作者单位】国防科技大学电子科学与工程学院,湖南长沙,410073;国防科技大学电子科学与工程学院,湖南长沙,410073;国防科技大学电子科学与工程学院,湖南长沙,410073
【正文语种】中文
【中图分类】TN958
【相关文献】
1.运动目标在阵列孔径UWB SAR图像上的表征及其在图像域GMTI方法中的应用 [J], 常玉林;黄晓涛;周红;周智敏
2.FPGA在SAR实时运动补偿系统中的应用 [J], 曲路正;李飞
3.SINS/GPS组合平滑估计在机载SAR实时运动补偿中的应用 [J], 宫晓琳;秦婷婷
4.机载干涉SAR运动补偿中地物目标定位误差的影响分析 [J], 李银伟;韦立登;向茂生
5.MEMS IMU随机误差建模在SAR运动补偿中的应用 [J], 庄晋升;汪丙南;向茂生因版权原因，仅展示原文概要，查看原文内容请购买。

A SENSOR FOR RADIANCE BASED DIAGNOSTICS

专利名称：A SENSOR FOR RADIANCE BASED DIAGNOSTICS发明人：SARUSSI, Israel申请号：EP99972083.2申请日：19991115公开号：EP1139865B1公开日：20071003专利内容由知识产权出版社提供摘要：The present invention provides a sensor (20) for radiance based diagnostics of body tissue including a performing component (21) and an adhering component (23). The performing component includes at least one radiance source (22) for radiating a tissue, and at least one detector (24) for detecting rays emitted from the radiance source. The adhering component is capable of fastening the performing component to a tissue such that the radiance source, and detector are facing and contiguous with the tissue. When operative, the adhering component fastens the performing component to the tissue to the extent that the detector receives only rays which are transmitted through or reflected from the tissue. The sensor may include a controlling device which senses external conditions and which is capable of controlling the sensor in response to the sensed conditions. The sensor may also be in communication with an electronic circuit, which may control the radiance source, and/or detector, performing analysis of data received by the detector.申请人：S P O MEDICAL EQUIPMENT LTD地址：IL国籍：IL代理机构：Vossius & Partner 更多信息请下载全文后查看。

1、下载文档前请自行甄别文档内容的完整性，平台不提供额外的编辑、内容补充、找答案等附加服务。
2、"仅部分预览"的文档,不可在线预览部分如存在完整性等问题,可反馈申请退款(可完整预览的文档不适用该条件!)。
3、如文档侵犯您的权益，请联系客服反馈,我们会尽快为您处理(人工客服工作时间：9:00-18:30)。

Index Terms—Motion compensation, simulation, synthetic aperture radar (SAR).
Manuscript received September 9, 2005; revised November 16, 2005. This work was supported in part by the Campania Region “Centro Regionale di Competenza su Analisi e Monitoraggio del Rischio Ambientale” and “Centro Regionale di Competenza su Tecnologie dell’Informazione e della Comunicazione.”
A. Iodice and D. Riccio are with the Dipartimento di Ingegneria Elettronica e delle Telecomunicazioni, Università di Napoli “Federico II,” 80125 Naples, Italy (e-mail: iodice@unina.it; daniele.riccio@unina.it).
Digital Object Identiﬁer 10.110DUCTION
S YNTHETIC aperture radar (SAR) is a remote sensing technique that allows the generation of microwave images of the Earth’s surface, independent of weather condition and sun illumination. SAR sensors can be either mounted on spacecrafts (spaceborne systems) or placed on airplanes (airborne systems).
In order to obtain the ﬁnal high-resolution image, the SAR raw signal received by the sensor must be properly processed, and standard SAR processing techniques are based on the assumption that the sensor ﬂight path is a straight line [1]. This is usually a good approximation for a short portion of the elliptical orbit of a spaceborne sensor, whereas in the case of airborne sensors signiﬁcant deviations from the ideal trajectory may occur: the effects of these deviations are compensated for by including in the processing chain a so-called motion compensation (MOCO) algorithm [2]–[6]. In order to easily test the effectiveness of such algorithms, a SAR raw signal simulation is very useful. To this aim, simple point-target simulators (i.e., simulators able to generate the raw signal produced by a single point-like scatterer) are usually employed, but it would certainly be useful to verify the effects of processing inaccuracies on perfectly known simulated extended scenes. In fact, these effects could be partly “masked” or “emphasized” when complex targets are considered, so that their impacts are different for different scenes or applications. In addition, an extended scene SAR raw signal simulator can be useful to help mission planning and to verify the impact of different system design choices on the ﬁnal image for different kinds of imaged scenes.
Giorgio Franceschetti, Life Fellow, IEEE, Antonio Iodice, Senior Member, IEEE, Stefano Perna, Student Member, IEEE, and Daniele Riccio, Senior Member, IEEE
S. Perna is with the Dipartimento di Ingegneria Elettronica e delle Telecomunicazioni, Università di Napoli “Federico II,” 80125 Naples, Italy, and also with the Istituto per il Rilevamento Elettromagnetico dell’Ambiente, 80124 Naples, Italy (e-mail: steperna@unina.it).
Abstract—Synthetic aperture radar (SAR) raw signal simulation is a useful tool for SAR system design, mission planning, processing algorithm testing, and inversion algorithm design. A two-dimensional (2-D) Fourier domain SAR raw signal simulator, exploiting the efﬁciency of fast Fourier transform algorithms, has been presented some years ago and is able to generate the raw signal corresponding to extended scenes. However, it cannot account for the effects of sensor trajectory deviations with respect to the nominal straight-line path. This paper explores the possibility of extending the efﬁcient Fourier domain simulation approach to the case of sensor trajectory deviations, which is more realistic for airborne SAR systems. We ﬁrst of all obtain a general and compact Fourier domain formulation of the SAR raw signal in the presence of arbitrary trajectory deviations, and show that in this general case no efﬁcient simulation scheme can be devised. However, we demonstrate that, if a narrow beam and slow trajectory deviation assumption is made, a full 2-D Fourier domain simulation can be used. This approach can be applied only to some SAR systems and/or trajectory deviations, but it has the advantage that processing time is practically not increased with respect to the nominal trajectory case. The validity limits of the approach are analytically evaluated. Some simulation results are ﬁnally presented in order to verify the effectiveness of the proposed simulation scheme. In another paper, which is the second part of this work, it will be shown that the narrow beam–slow deviation assumption can be relaxed, at the expense of computation efﬁciency, if a one-dimensional azimuth Fourier domain processing followed by a range time-domain integration is used.