应用于雷达方位超分辨的改进Richardson-Lucy算法

An Improved Richardson-Lucy Algorithm Used in Radar Azimuth Super-resolution

方位超分辨是国内外雷达界持续探索的一项技术难题。为解决高斯噪声情况下天线低通效应造成的方位低分辨力问题,利用基于泊松噪声情况下Richardson-Lucy(RL)算法对其进行研究。针对RL算法中高斯噪声被放大出现虚假目标和RL算法收敛速度慢的缺点,提出了一种改进RL算法。该算法首先使用低通滤波去除高频段的噪声,然后利用加速RL算法进行方位超分辨。仿真结果表明:与原RL算法相比,改进RL算法能降低高斯噪声的影响,消除虚假目标;在信噪比低至0 dB时,相对于半功率波束宽度,改进RL算法的最大分辩倍数为1.8倍,与维纳逆滤波算法相比具有较强的噪声适应能力,可以应用于雷达方位超分辨。

Azimuth super-resolution remains a technical problem needed to be explored in radar domain both at home and abroad. In order to solve the low resolution problem caused by antenna’s low-pass effect under the condition of Gaussian noise, Richardson-Lucy （ RL） algorithm is used in research based on the Poisson noise. For the drawback of false target because of amplified effect of Gaussian noise and slow con-vergence rate, an improved RL algorithm is proposed. Firstly, the improved algorithm uses the low-pass filter to remove the noise in high frequency band, then its azimuth super resolution is solved by the acceler-ated RL algorithm. Simulation results show that, compared with the original RL algorithm, the improved RL algorithm can reduce the Gaussian noise, eliminate false targets, with regards to the half-power beam-width, its maximum resolution ratio can be reach 1. 8 times when the signal-to-noise ratio（SNR） is low to 0 dB, which shows the improved RL algorithm has strong noise adaptation capability compared with Wie-ner inverse filtering algorithm and can be used in radar azimuth super-resolution.