无源毫米波实时成像PNL超分辨算法  

PNL super-resolution algorithm for PMMW imaging in real-time

作　　者：姜正茂[1] 杨建宇[1] 李良超[1] 郑鑫[1] 

机构地区：[1]电子科技大学电子工程学院,四川成都610054

出　　处：《系统工程与电子技术》2012年第1期7-11,共5页Systems Engineering and Electronics

基　　金：国家自然科学基金(60632020)资助课题

摘　　要：投影Landweber(projected Landweber,PL)算法具有良好的频谱外推能力,能够应用于低信噪比降晰图像,是一种适合无源毫米波成像的超分辨算法。但其缺点是收敛速度缓慢,运算量不稳定,难以满足实时性要求。针对实时性问题,提出一种投影Newton-Landweber(projected Newton-Landweber,PNL)超分辨算法,首先使用Newton求逆法得到粗恢复图像,然后运用PL算法对图像做精细恢复。实验结果表明,该算法显著提高了收敛速度,图像恢复质量接近PL算法的性能。The projected Landweber（PL） algorithm is competent for passive millimeter wave（PMMW） imaging for its spectrum extrapolation performance and low signal to noise ratio images processing ability,however,the slow speed of convergence and the computation instability limit its applications in real-time.A projected Newton-Landweber（PNL） algorithm is proposed to achieve real-time ability on the basis of the PL algorithm.Firstly,a fast Newton inversion algorithm is introduced to get coarse image recovery.Then the PL algorithm is used to get refined image recovery.Experimental results demonstrate that the performance of the PNL algorithm approaches that of the PL algorithm,moreover the computation is stable and far less than the PL algorithm.

关 键 词：无源毫米波成像 超分辨 投影Landweber Newton求逆 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]