基于稀疏分解的SAR射频干扰抑制与回波重构技术

Ultra Wide Band (UWB) Synthetic Aperture Radar (SAR) owns an ability to detect the targets shaded in the foliage or under the ground. But the television, radio and communication signals which operating at VHF/UHF band bring severe Radio Frequency Interferences （RFI） to UWB SAR. Since the space-varying and time-varying components of RFI are inevitable for a wide band radar, the non-stationary RFI can not be suppressed effectively by the Current RFI suppression algorithms based on a stationary model. We propose a non-stationary RFI suppression and target echo reconstruction technique based on sparse decomposition to solve the problem in this subject. Considering the sparsity of significant RFI in time-frequency plane，the time-frequency analysis technique is used to research the characteristics of the RFI in UWB SAR. A redundant dictionary which can match the local characteristics of non-stationary RFI well will be constructed to replace the traditional standard orthogonal basis. Then sparse decomposition technique is used to reconstruct RFI. After eliminate the reconstructed RFI, the target echo would have more or less distortion. Therefore it is necessary to utilize prior information, such as the characteristics of transmitting waveform and the sparsity of targets，to reconstruct target echo and improve the imaging quality. In order to improve computation efficiency and reduce atoms storage space, this subject will focus on a more efficient algorithm for sparse decomposition. To guarantee the robustness and the efficiency of the reconstruction algorithm，the subject will take different levels of noise into account and apply the algorithm to area target SAR imaging. The sparse decomposition method based on some time-frequency dictionaries has a relatively exact representation for RFI including the time-varying components. The reconstruction may suppress the high side-lobe caused by the traditional estimation method. The research will make a breakthrough in RFI suppression of UWB SAR.

超宽带合成孔径雷达（UWB SAR）具备强的穿透探测隐蔽目标能力，但同频段电视、广播和通信信号对雷达造成严重射频干扰（RFI），且干扰具有时变和空变性。以干扰近似平稳为基础发展起来的传统RFI抑制技术不能有效解决非平稳RFI抑制问题。 为此，本项目提出基于稀疏分解的非平稳RFI抑制和回波重构技术，技术途径为：首先采用时频分析技术研究UWB SAR射频干扰特征，提出与干扰相匹配的冗余字典代替传统正交基，然后基于稀疏分解算法估计RFI干扰，对去除RFI后的回波基于信号和目标特征进行重构，提高成像质量。项目需重点研究稀疏分解的快速算法和RFI抑制技术的面目标成像适用性问题，提高RFI估计和回波重构的计算效率，增强算法对含噪数据的稳健性。时频冗余字典能实现对包括时变干扰在内的射频RFI的准确表征，回波重构技术可解决传统RFI抑制技术的高旁瓣问题，研究有望在RFI抑制问题上取得突破。

超宽带合成孔径雷达（UWB SAR）具备树林和地下隐蔽目标穿透探测能力，但同频段电视、广播和通信信号对雷达造成严重射频干扰（RFI），且干扰具有时变性和空变性。本项目从实测回波和RFI信号特征入手，研究基于稀疏分解的非平稳RFI抑制和回波重构技术，项目研究成果包括：1、采用时频分析技术研究实测回波中的射频干扰特征，讨论了VHF/UHF频段RFI抑制的难点。2、研究了基于时频分析和稀疏分解的RFI抑制技术，利用实测数据验证了基于特征子空间和短时傅里叶变换的RFI抑制算法性能。研究了基于Gabor过完备字典和FFT正交基的稀疏分解RFI抑制算法，实测数据验证了成像的目标失真更小但运算量增大，为此我们基于实测RFI的时域稳定度改进了字典结构，提高了算法效率。3、提出了一种改进的强散射点剔除方法提高窄带RFI自适应滤波器对时变RFI抑制效果，可大大减小传统算法的距离旁瓣效应；针对非参数法与自适应算法在RFI抑制上的互补性，提出通过图像域对消技术抑制幅度图上的残余RFI杂波；4、研究了基于独立分量分析的杂波抑制技术，提出了一种利用MIMO多次快拍数据实现杂波抑制的算法;研究了基于压缩感知的RFI抑制和目标重构算法框架，提出了一种二阶OMP的RFI抑制算法提高效率。基于项目研究成果，课题组研究人员在国内外期刊和国际会议共发表学术论文12篇，其中SCI检索1篇，EI检索6篇，申请国家发明专利3项。