基于压缩感知的双站ISAR成像研究

发布时间：2018-03-14 05:32

本文选题：压缩感知　切入点：双站ISAR　出处：《电子科技大学》2014年硕士论文　论文类型：学位论文

【摘要】：双站逆合成孔径雷达(双站ISAR)相较于单站逆合成孔径雷达其作用距离更远,而且由于双站ISAR的收发雷达分置,它的抗干扰能力、抗截获能力等安全性能也更为出色,因此双站ISAR具有较高的民用及军用价值。但是正是由于收发雷达分置以及双站夹角,双站ISAR成像的分辨率低于相应的单站ISAR,且双站ISAR目标通常为非合作目标,其运动规律性未知,导致接收端雷达可能收到缺失不完整的回波信号,如果对此回波采用传统的奈奎斯特采样方法,无法得到无失真的目标图像,传统的解决方法是增大发射雷达信号的带宽来提高距离分辨率,而在信号缺失部分利用线性插值或者全极点模型匹配法,但是这两种方法无法在双站夹角过大或者信号稀疏度过高时使成像结果达到理想分辨率。压缩感知(Compressive Sensing,CS)理论指出,在对稀疏信号进行采样时,可以对该信号进行远低于奈奎斯特采样率的压缩采样得到少量测量值,再采取合适的优化算法来重构原始信号。本文将压缩感知技术应用于双站ISAR成像中,针对实际情况中大双站夹角与稀疏孔径带来的成像低分辩问题,围绕相关重构理论,从以下几个方面展开了研究:(1)由传统信号采样理论的缺点引出压缩感知理论,着重阐述压缩感知技术中的三个最关键步骤:信号的稀疏表示、观测矩阵的构造以及信号的重构算法。为后文基于压缩感知的双站ISAR成像算法奠定理论基础。(2)介绍传统的双站ISAR成像算法之一RD算法,分析该算法的分辨率。着重分析双站夹角大小对分辨率的影响、稀疏孔径对分辨率的影响,从而引出后文将压缩感知理论应用于大双站夹角ISAR成像中。(3)把压缩感知理论应用到双站ISAR成像算法中。先分析转台成像模型并用步进频率信号作为发射信号得到目标回波,再根据回波构造稀疏基以得到回波的稀疏表示,然后构造适当的观测矩阵,把信号投影到更低维的空间上得到观测样本,最后将问题转化成从观测样本中重构原始信号。在重构信号的问题上,利用贪婪算法和基于范数最小化的凸优化算法来解决。在理论研究后,通过仿真实验对各算法成像的有效性进行了验证与分析。并对比了基于凸优化算法的GPSR重构算法、BP重构算法和基于贪婪算法的OMP重构算法以及GAPES重构算法。(4)通过仿真实验从重构误差、抗噪性能对各种重构算法进行了对比分析。
[Abstract]:Bistatic inverse synthetic Aperture Radar (bistatic synthetic Aperture Radar) has a longer operating range than the single station inverse synthetic Aperture Radar (ISAR), and its anti-jamming ability and anti-interception capability are also better because of the bistatic ISAR's transceiver and transceiver. Therefore, bistatic ISAR is of high civilian and military value, but the resolution of bistatic ISAR imaging is lower than that of the corresponding mono-station ISAR because of the transceiver radar separation and bistatic angle, and the bistatic ISAR target is usually a non-cooperative target. The unknown regularity of its motion may result in the radar receiving incomplete missing echo signal. If the traditional Nyquist sampling method is used, the target image can not be obtained without distortion. The traditional solution is to increase the bandwidth of the transmitted radar signal to improve the range resolution, and the linear interpolation or full pole model matching method is used in the missing part of the signal. However, these two methods can not achieve the ideal resolution when the bistatic angle is too large or the signal is sparse. A small amount of measured data can be obtained by compressed sampling with a much lower sampling rate than Nyquist sampling rate, and then a suitable optimization algorithm is adopted to reconstruct the original signal. In this paper, compression sensing technique is applied to bistatic ISAR imaging. In order to solve the problem of low resolution caused by large bistatic angle and sparse aperture, the compression sensing theory is derived from the shortcomings of traditional signal sampling theory. In this paper, three key steps in compression sensing technology, namely sparse representation of signals, are discussed. The construction of observation matrix and the reconstruction algorithm of signal are introduced. Rd algorithm, one of the traditional bistatic ISAR imaging algorithms, is introduced in this paper, which lays a theoretical foundation for the bistatic ISAR imaging algorithm based on compression perception. The resolution of the algorithm is analyzed. The influence of bistatic angle size on the resolution and the effect of sparse aperture on the resolution are analyzed. Then the compression sensing theory is applied to the large bistatic angle ISAR imaging. The compression sensing theory is applied to the bistatic ISAR imaging algorithm. Firstly, the imaging model of the turntable is analyzed and the step frequency signal is used as the transmitting signal to get the target echo. Then the sparse basis is constructed according to the echo to obtain the sparse representation of the echo, and then an appropriate observation matrix is constructed to project the signal onto a lower dimensional space to obtain the observation sample. Finally, the problem is transformed to reconstruct the original signal from the observation sample. The greedy algorithm and the convex optimization algorithm based on norm minimization are used to solve the problem. The validity of each algorithm is verified and analyzed by simulation experiments, and the GPSR reconstruction algorithm based on convex optimization algorithm is compared with the OMP reconstruction algorithm based on greedy algorithm and the GAPES reconstruction algorithm. Simulation experiment from reconstruction error, The anti-noise performance is compared and analyzed for various reconstruction algorithms.
【学位授予单位】：电子科技大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TN957.52

【相似文献】