摘要
随着成像雷达图像分辨率的不断提升,对分布式目标的探测性能指标要求越来越高,然而现有的恒虚警检测算法尚未形成适用于多分辨单元的目标检测性能评估理论体系。提出一种分布式目标滑窗检测方法。首先,基于待检测目标RCS(radar cross section,RCS)、平面尺寸以及雷达二维分辨率等先验信息对二维滑窗进行设计,实现了任意航向目标的最优匹配。接着,利用M/N准则对滑窗内目标占据的多分辨单元回波进行直接处理,避免了人为构建扩展检测量引起的检测性能损失。最后,从系统虚警指标要求出发,通过挖掘分布式目标与所占据单个分辨单元的检测性能关系,设计了一套分布式目标探测性能的定量评估方法,可在满足虚警性能前提下大幅提升分布式目标的检测概率。该算法不需要人为进行复杂的扩展检测量设计,为高分辨图像目标稳健检测和性能评估开辟了一种新途径。
With the continuous improvement of radar resolution,a higher detection performance requirement is put forward for distributed targets.However,the existing algorithms have not yet formed a detection performance evaluation theory.In this paper,a novel sliding window detection method for distributed targets is presented.Firstly,the two-dimensional sliding window is designed to achieve the optimal matching for the specified target with any navigation direction based on the prior information,i.e.,RCS,target size and radar resolution.Then,the target units within the sliding window are processed by means of the criterion,and thus to avoid the detection performance loss caused by the artificial construction of extended detection factor.Finally,by mining the detection performance relationship between the distributed target and that of a single resolution unit,a new quantitative evaluation method with regard to high-resolution radar is designed from the perspective of the given system false alarm index,where the detection probability of distributed target on the premise of satisfying the false alarm performance can be greatly improved.The proposed algorithm does not need to design the complex extended detection factor,which provides a new way for robust detection and performance evaluation for distributed targets.
作者
李渝
吴涛
杨文海
张军
雷红文
李财品
段崇棣
LI Yu;WU Tao;YANG Wenhai;ZHANG Jun;LEI Hongwen;LI Caipin;DUAN Chongdi(China Academy of Space Technology(Xi’an),Xi’an 710000,China)
出处
《空间电子技术》
2022年第2期64-68,共5页
Space Electronic Technology
基金
国家自然科学基金青年基金(编号:62107033,62001374)。
关键词
分布式目标
滑窗
二维分辨率
检测概率
distributed target
sliding window
two-dimensional resolution
detection probability
