一种超宽带随机噪声压缩感知雷达成像方法

An imaging method of ultrawideband random noise compressive sensing radar

超宽带随机噪声雷达具有较高的成像分辨率和较强的抗电磁干扰能力,然而其高带宽特性给信号的采集硬件带来了沉重负担,压缩感知理论的出现可以很好地解决该问题。结合合成孔径雷达成像原理,提出一种适用于超宽带随机噪声雷达的压缩感知成像算法,该算法不仅可以极大地降低系统的采样频率并且可以使用较少的数据集对场景成像,因此可以节省系统的存储空间。成像结果表明,这种基于压缩感知的成像算法,在利用较少的测量数据集时仍然具有较好的成像效果,而采用传统线性成像方法会产生严重的伪影。由于压缩感知重建算法的计算复杂性随着成像维度的增加而增大,基于压缩感知的方法在时间和内存消耗方面远高于传统线性方法,以至于无法获得大尺度场景的重建,因此采用一种近似的重构算法,该算法在极大地降低计算复杂性的同时,保持着较好的计算精度,可以实现对大尺度场景的成像,文中实现了点数为2048×2048的场景重建。

The ultrawideband random noise radar has high imaging resolution and strong anti⁃electromagnetic interference abilities.However,the characteristic of its large bandwidth brings a heavy burden to the signal acquisition equipment.The emergence of compressive sensing(CS)theory is promising to solve this problem perfectly.In combination with the principles of synthetic aperture radar(SAR),a CS⁃based imaging algorithm suitable for ultrawideband random noise radar is presented.It can greatly reduce the sampling frequency of the hardware and achieve the scene imaging by means of fewer datasets,so it can save storage space of the system.The imaging results show this CS⁃based imaging algorithm has a good imaging effect with fewer measuring datasets,while the traditional linear imaging method encounters a severe artifact effect.Since that the computational complexity of CS reconstruction algorithm is increased with the increment of the imaging dimension,the time and memory consumption of CS⁃based method is much higher than that of the traditional linear method,making it impossible to reconstruct large⁃scale scenes.To solve this problem,an approximate reconstruction algorithm is presented.It can greatly reduce the computational complexity while maintaining a high accuracy,so the large⁃scale scene imaging can be realized.For example,the reconstruction of the 2048×2048 points sense is achieved with the proposed method.