信号处理2026，Vol.42Issue(6)：783-796,14.DOI:10.12466/xhcl.2026.06.002
基于贪婪积累的信息几何雷达弱小目标检测前跟踪方法
Greedy Integration-Based Information Geometry Track-Before-Detect Method for Weak Radar Targets
摘要
Abstract
This study addresses the difficulty of detecting weak radar targets in complex clutter backgrounds.The supe-rior capability of information geometry detection methods in distinguishing targets from clutter on manifolds and the multi-frame information integration ability of track-before-detect(TBD)algorithms are leveraged.A greedy integration-based information geometry TBD method is proposed.By establishing a high-dimensional manifold representation of ra-dar multi-frame echo signals,the distance between the target and clutter on the manifold is used as a feature difference.Through the integration of multi-frame feature differences,the target and clutter feature trajectories on the multi-frame manifold gradually separate,significantly increasing their dissimilarity.This method uses feature differences on the manifold as multi-frame integration quantities and uses a greedy integration algorithm to achieve multi-frame informa-tion integration.This distinguishes targets from complex clutter backgrounds and outputs target trajectories.Through a series of locally optimal forward decisions made by the greedy integration algorithm,multi-frame joint processing is achieved while outputting the current best estimate frame-by-frame.Experimental validation using measured data demon-strates that the greedy integration-based information geometry TBD algorithm achieves superior detection performance compared with traditional energy-feature-based TBD algorithms.The proposed method can achieve a detection perfor-mance like that of the matrix information geometry dynamic programming TBD algorithm.It also reduces the computa-tion time by 80%.Compared to traditional energy-based TBD algorithms,the proposed method improves the detection signal-to-clutter ratio by approximately 2 dB,increases the detection probability by more than 10%,and achieves a higher trajectory estimation accuracy.关键词
检测前跟踪/信息几何/贪婪积累/雷达目标检测Key words
track-before-detect/information geometry/greedy integration/radar target detection分类
信息技术与安全科学引用本文复制引用
刘金国,吴昊,程永强,杨政,华小强..基于贪婪积累的信息几何雷达弱小目标检测前跟踪方法[J].信号处理,2026,42(6):783-796,14.基金项目
国家自然科学基金(62322122,62371458,62301598,62401583) The National Natural Science Foundation of China(62322122,62371458,62301598,62401583) （62322122,62371458,62301598,62401583）
