用于雷达方位超分辨的范数正则化方法  

Norm regularization method for radar azimuth super-resolution

作　　者：邹建武[1] 祝明波[1] 高明哲[1] 李相平[1] 

机构地区：[1]海军航空工程学院电子信息工程系,山东烟台264001

出　　处：《系统工程与电子技术》2014年第8期1500-1504,共5页Systems Engineering and Electronics

基　　金："泰山学者"建设工程专项经费(ts20081130)资助课题

摘　　要：方位超分辨是国内外雷达界长期以来持续探索的一项技术难题,针对求解过程中所遇到的病态性,本文深入研究了范数正则化方法。利用L2范数正则化方法对方位超分辨模型进行求解,针对L2范数正则化方法的不足并考虑到目标信号的稀疏性质,建立了L1范数正则化模型。为提高求解的计算效率,将其转化为半定规划模型,用预测校正原对偶路径跟踪法进行求解。针对不同信噪比情况进行了计算机仿真,初步结果表明,两种方法都能用于雷达方位超分辨,且在相同条件下L1范数正则化方法分辨性能更好,具有较强的噪声适应能力,在信噪比低至0dB时,分辨力提高1.7倍。The azimuth super-resolution is a technique problem which has been explored continuously in the field of radar.The L 2 norm regularization method is used for the solution to radar azimuth super-resolution problem,because of the ill-condition encountered in the solution process.The L 1 norm regularization model is also established for the shortage of the L 2 norm regularization method and the sparseness of the target signal.In order to improve the computational efficiency,the L 1 norm regularization model is transformed into a semi-defi-nite programming model which is solved by the predictor-corrector primal-dual path-following method.The computer simulation for different signal-to-noise ratio （SNR）is made.The preliminary findings show that both methods can be used in radar azimuth super-resolution,and the performance of the L1 norm regularization method is better under the same conditions,with a strong noise adaptive ability.It shows when SNR is not less than 0 dB,the resolution is increased by 1.7 times.

关 键 词：超分辨 范数正则化 半定规划 稀疏 

分 类 号：TN959[电子电信—信号与信息处理]