摘要
针对毫米波雷达由于阵列分辨率限制等因素,其点云成像存在点云密度稀疏、精度低等问题,提出了一种基于合成孔径雷达(SAR)技术的车载平台毫米波雷达三维(距离、方位、俯仰)高分辨点云视频成像方法。首先,利用时域后向投影(BP)成像算法解决近距宽角域成像聚焦难题,获得高分辨二维视频SAR图像;然后,通过基于幅度阈值的复交替方向乘子法压缩感知网络(CV-ADMM-CSNet)获得场景高程信息,通过模型化方法并结合数据训练,实现快速实时高分辨三维成像;最后,结合多帧视频成像处理获得动态三维高分辨点云图像。仿真和实测数据实验,验证了本文算法的有效性。
Due to the limitations of array resolution and etc,the point cloud imaging technology of millimeter-wave radar has the disadvantages of sparse distribution of point cloud and low-precision location.In this paper,a novel point cloud video imaging algorithm of 3D millimeter-wave radar on vehicle platform using the synthetic aperture radar(SAR)technology is proposed.First,the time-domain back-projection(BP)algorithm is applied to solve the slant-range wide-angle imaging focusing problem,obtaining well-focused high-resolution 2D image.Second,the amplitude threshold-based Complex-Valued Alternating Direction Method of Multipliers Compressed Sensing Net(CV-ADMM-CSNet)is applied to obtain elevation information of scenes,which achieves fast and real-time high-resolution 3D imaging through modeling methods combined with data training.Finally,the dynamic 3D point cloud image can be successfully generated in combination with the multi-frame SAR imaging.The experimental analysis using simulated and measured data is performed to confirm the effectiveness of the proposed algorithm.
作者
崔硕
蒋梦杰
张邦杰
陈宇智
徐刚
CUI Shuo;JIANG Mengjie;ZHANG Bangjie;CHEN Yuzhi;XU Gang(State Key Laboratory of Millimeter Waves,Southeast University,Nanjing Jiangsu 210096,China)
出处
《现代雷达》
CSCD
北大核心
2024年第10期32-38,共7页
Modern Radar
基金
中央高校基本科研业务费资助项目(2242022K60008)。
关键词
毫米波雷达
三维点云视频成像
合成孔径雷达
millimeter-wave radar
3D point cloud video imaging
synthetic aperture radar
作者简介
崔硕,男,2002年生,博士研究生,研究方向为合成孔径雷达成像;蒋梦杰,男,1995年生,博士研究生,研究方向为毫米波雷达系统及合成孔径雷达成像;张邦杰,男,1995年生,博士研究生,研究方向为毫米波雷达系统、合成孔径雷达层析成像和逆合成孔径雷达成像;陈宇智,男,1998年生,硕士研究生,研究方向为合成孔径雷达成像;徐刚,男,1987年生,博士,教授,博士生导师,研究方向为SAR/ISAR成像和毫米波雷达。
