| 摘要: |
| 本文针对低空小型无人机在雷达探测中散射截面积小、相干积累时间短等问题，提出一种基于贝叶斯统计机器学习的逆合成孔径雷达超分辨成像方法。利用无人机相对空域背景的稀疏性先验知识引入重尾的拉普拉斯先验概率分布，并基于观测系统噪声高斯分布假设建立贝叶斯后验推理模型。针对先验分布的非共轭性，引入分层贝叶斯模型。最后应用变分贝叶斯期望最大算法，解析求解目标后向散射系数后验概率密度函数，并校正目标非系统性平动误差及其造成的成像散焦。与传统方法相比，该方法能够有效解决无人机目标雷达散射截面积较小带来的成像信噪比低以及相干积累时间较短带来的成像分辨率低等问题。仿真实验结果证明了本文所提方法的有效性和优越性。 |
| 关键词: 无人机 逆合成孔径雷达 贝叶斯学习 机器学习 压缩感知 |
| DOI：DOI:10.3969/j.issn.1672-2337.2020.03.011 |
| 分类号:TN958 |
| 基金项目:国家自然科学基金（No.61601470）； 天津市大学生创新创业项目（No.2050205） |
|
| Low-Altitude and Small-Size UAV ISAR Imagery Based on Sparse Bayesian Learning |
|
LIU Minghao， XU Jiu， ZHAO Fuchenglong， CHENG Kaifei， YANG Lei
|
|
College of Electronic Information and Automation, Civil Aviation University of China, Tianjin300300, China
|
| Abstract: |
| Aiming at the problems of small radar cross-section and short coherent processing interval in radar detection of low-altitude and small-size UAV, an inverse synthetic aperture radar (ISAR) super-resolution imaging method based on Bayesian statistical machine learning is proposed in this paper. The sparse prior characteristics of UAV with respect to the airspace background is utilized by introducing a heavy-tailed Laplacian prior probability distribution, and a Bayesian posterior inference model is established based on the Gaussian distribution assumption of observation system noise. In view of the non-conjugation of prior distribution, a hierarchical Bayesian model is introduced. The variational Bayesian expectation maximization algorithm is applied to the hierarchical Bayesian model, and the posterior probability density function of the target is solved analytically. Meanwhile, the imaging defocus caused by the non-systematic translational error of the backscatter coefficient of target is corrected. Compared with traditional methods, this method can effectively solve the problems of the low imaging signal-to-noise ratio caused by the small RCS of UAV target and the low imaging resolution caused by the short coherent processing interval. The experimental results of simulation prove the effectiveness and superiority of the proposed method. |
| Key words: unmanned aerial vehicle inverse synthetic aperture radar Bayesian learning machine learning compressed sensing |
