【作者】 卢光跃；

【导师】 保铮；

【作者基本信息】 西安电子科技大学 ， 信号与信息处理， 1999， 博士

【摘要】 逆合成孔径雷达（ISAR）技术通常是雷达不动，而对非合作性运动目标进行成像，它作为对运动目标成像的关键技术，具有远距离、全天时和全天候的优点，在民用和军事上的广泛应用前景而倍受关注。 常规ISAR成像的研究均基于散射点模型，利用简单的距离多普勒算法（RDA）进行成像的，由于实际ISAR目标为非合作性的，目标的运动相当复杂，因此，简易的成像技术在对实际数据处理时，不总能得到满意的成像结果。为此，本文的主要工作一方面对实测数据中存在的问题如包络对齐、散射点越距离单元走动的问题进行了进一步研究，提出了一些较为有效的算法；另一方面，在ISAR实用化方面进行了一定的研究，提出了一种ISAR机动目标实时成像的简化算法。具体内容如下： 针对传统的包络相关对齐法在对齐处理时存在的漂移和突跳，西电雷达成像小组提出了非参数化的包络最小熵对齐方法。在第二章中，我们对该方法的适用性进行理论分析，理论分析的结果表明，最小熵准则的包络对齐方法适宜于对平稳飞行和机动飞行目标进行包络对齐，同时给出实用的最小熵包络对齐方法，多批实测数据的处理结果表明对非参数化的最小熵方法理论分析的正确性及算法本身的有效性。 由于常规包络对齐，包括相关法及最小熵方法，均基于目标为刚体进行的，然而，当目标上包含游动部件（如An-26飞机上的螺旋桨）时，直接利用常规包络对齐方法通常得不到满意的结果。针对这一问题，第三章结合实测的螺旋桨飞机（An-26）数据，分析螺旋桨对目标回波及包络对齐的影响，定义了回波的动态范围，利用回波的动态范围进行自适应稳定回波的选取以形成反映目标稳定部件反射特性的参考信号，并通过自适应门限选取以弱化螺旋桨子回波对包络对齐的影响，对实际测量数据处理的结果表明本章对包络对齐方法的改进是有效的。 散射点在纵向的位移是产生多普勒频移的基础，然而，散射点在纵向的位移也是散射点越分辨单元走动（MTRC）的根源。第四章根据ISAR成像的具体特点，即目标为远距离、小目标，给出了散射点MTRC与其坐标的关系，进而将二维MTRC校正问题降维为两个一维校正问题，简化了计算，并分别给出了散射点越距离分辨单元/多普勒分辨单元走动的补偿方法，对匀速转动目标而言，该算法不需要目标的转动信息即可进行散射点越距离单元走动的校正，适宜于一对运动参数未知B 目标的情况。 第五章探讨了上一章提出的MltC校正算法与简易的RDA、极坐标插值算法oFA）在处理MTRC方面的差异，并从算法的模糊半径对三种算法进行比较；最后，利用仿真数据及微波暗室数据班bflltC 校正算法的有效性，舰分析的结果也表明MTRC校正算法能够有效地扩大成像范围，能满足对B 大目标进行成像的需要。 由于ISAR成像实际上是三维目标在二维成像平面上的投影，当目标机动飞行时，目标的成像投影平面随时间变化，用常规的RD算法得到的目标图像将会发生严重模糊，为此西电雷达成像小组提出距离一瞬时多普勒（RI）算法进行机动目标成像。然而，在ISAR技术逐步实用化的过程中，需要研究适宜于实时成像的脑算法，为此，第六章翻了一种基于信号分量自适应的STFT方法，该方法保定信号分量在 战平稳长度内为一单频率分量，尽可能地利用该分量的数据长度，同时利用‘洁净’技术逐个估计各频率分量。由于该技术利用n可和‘洁净’思想，其计算量较小，但该方法的近似性也是显而易见的；于是，将散射点回波近似为线性调频信号，并结合ISAR ＄t中的特殊性（如测点俄离单元走动）的情况下，提出基于自适应LFM信号分量的‘洁净’技术估计回波的瞬时频率，从而得到目标高分赠的距离口时多普勒图像；对实验ISAN数据的处理结果表明该方法的有效性。 第七章对全文工作进行了总结，并简单探讨了ISAR成像未来的研究方向及存在的问题。更多还原

【Abstract】 ANSmCTInverse Synthetic APerture sadar (ISAR), in which the radar isstationary and the target is moving, can work a1l day, al1 weather andhas become one of the nain techniqUes in imaging the moving target farfrom the sensor. More and more research on ISAR has been carried outbecause of 1ts potentia1 app1ications in civi1 and mi1itary app1ications.A1most of a11, the conventional ISAR research 1s based on the scatteringmode1, that is, the scattering prOPerty of the target ca-n be modeled tyan asseInbly scatterers and the thege of the target can be achieved tysimple Range--DoPP1er a1gorithIn (RDA). Because of the non--cooperativityof the target and the cOmp1exity of the movenent, the ISAR techniqUes,inc1uding the ra-nge alignInent and theging a1goritbo, can not a1waysresult in satisfactory resu1t in rea1 data processing. The main work ofthis dissertation focuses on two aspects. The first,, the existingprob1ems, such as range a1irment and scatterer Migration ThroughResolution Cell (mRC), in real data processing, are discussed and thecorresponding a1gorithins are proposed to deal with the problems; Second,the real time imaging for maneuvering targets is carried out and a simp1eRange--Instantaneous DoPPler (RID) a1gorithIn is deve1oped. The detai1sof the dissertation is as fo1low:In ISAn imaging, data--driven coInPensation of motion is necessarybecause of the difficu1ty in measuring the motion paraneters of thenon--cooperative targets, where the range a1impent is one of the key steps.The conventional range a1ipent method, i. e., the enve1ope corre1ationmethod, often brings inaccurate aligrment resulting from drift errorsand jmp errors. To remove the errors, the non--paraneter minina entroPy(ME) method was proposed. As indicated in a 1ot of results of real data,ME method can guarantee the g1oba1 range alignent. In ChaPter 2, thetheoretica1 analysis of ME method is presented. The iraging resu1ts ofreal data verify the correctness of the ana1ysis and the effectivenessof the ME method.Because conventiona1 range a1igrment methods asscoe that scatterers’positions are a1most fixed during the iareging interval, which reqUiredthe target be rigid, it is not easy for us to obtain good resu1t for thetarget with moving P8rts (such as the prOPe1lers on An-26 aircraft) usingthe conventiona1 bethod. In Chapter 3, the effects of the prOPel1ers sub-echoes on the return and the rope alimpnt are ana1yzed using the rea1data of An-26 aircraft. A modified method for this specia1 case isproposed, in which several returns with sca11er dynanic range areselected first tc generate the reference echo, and then the effect ofthe prope1lers on the returns with 1arger dynanic range is weakenedthrough the adaptive thresho1ding. The fina1 results of the rea1 dataconfirn the rodified range alignInent method.After the motion cOmpensation, the image of the target can be obtainedby tuge-DoPP1er method. To obtain radar thege of a target, certain op1eof the target with respect to Radar Light Of Sight (RLOS) during thecoherent integration interva1 is reguired, that is, the motion is thesource of ISAR icagieq. however, it is also the source of problem (suchas rv). Actua11y, scatterers’ rm hapPens inevitab1y in higherreso1ution radar image,which wi11 resu1t in the POint--spread functionbroadeninq and image reso1ution decreasing. In ChaPter 4, the scatterers’migration in ISAR is analyzed and the 2--D space--variant comPensation isreduced Lo two l--D coInPensations. And a siInP1e and efficient method isproposed to cOInPensate MTR - In the proposed wrRC COInPensation AlgorithIn,scatterer migration range resolution ce1l can be cOInPensated withoutthe know1edge of rotating parameters. CoInPensation resu1ts of thesiamlated data and the rea1 data co11ected in a microwave anechoic chaInbershow that the reso1ution of the fina1 radar theges is twroved.The re1ationship between rv CoInPensation a1gorittri更多还原