传统的高分辨率雷达成像技术面临着需要大量数据的问题,压缩感知理论的应用可以避免这一问题的出现。在压缩感知理论的基础上进行改进,通过使用字典优化的压缩感知高分辨率雷达成像方法,即用复杂正弦信号来实现高精度重构算法,进而在有限数据情况下实现高分辨雷达成像。考虑到字典失配问题易对成像性能产生影响,使用稀疏表示将 复正弦信号转变成傅里叶字典,稀疏表示字典在信号重构过程中是可动态优化的,且其优化可通过频率格点调整来实现,通过使用变分期望(最大化算法)实现稀疏系数重构和频率的迭代更新来实现上述算法。在一组暗室所测数据上进行实验,实验结果表明,该算法可以有效解决字典失配问题对成像结果的影响,且能够实现利用有限数据生成高质量的高分辨距离像。

The traditional high-resolution radar imaging technology faces the problem of requiring large amounts of data. The application of compressed sensing theory can avoid this problem. Improved on the basis of compressed sensing theory, through the use of dictionary-optimized compressed sensing high-resolution radar imaging method, that is, using complex sinusoidal signals to achieve high-precision reconstruction algorithms, and then achieve high-resolution radar imaging with limited data. Considering that the problem of dictionary mismatch is likely to affect the imaging performance, the sparse representation is used to transform the complex sine signal into a Fourier dictionary. The sparse representation dictionary can be dynamically optimized during the signal reconstruction process, and its optimization can be achieved through frequency grid points. Adjustment is achieved. The article implements the above algorithm by using variation expectation (maximization algorithm) to achieve sparse coefficient reconstruction and iterative update of frequency. Experiments are conducted on a set of measured data in a darkroom. Through the study of sparse parameter recovery and imaging comparison methods, the experimental results show that the proposed algorithm can effectively solve the impact of dictionary mismatch problem on the imaging results, and can realize the use of limited data to generate high The highresolution range profile of the quality shows the effectiveness of the proposed method.

TN957.51

宁夏师范学院校级重点科研基金资助项目(SFZD2019-19)