项目名称: 基于多贝努利随机集的弱小目标TBD方法研究
项目编号: No.61301289
项目类型: 青年科学基金项目
立项/批准年度: 2014
项目学科: 无线电电子学、电信技术
项目作者: 李翠芸
作者单位: 西安电子科技大学
项目金额: 24万元
中文摘要: 低信噪比下红外弱小目标的检测,由于其信号能量低、背景复杂、目标数时变等特点,一直是国内外相关研究领域的一个热点与难点问题。本项目以随机有限集为理论基础,结合非平稳信号处理方法,拟研究基于多贝努利随机集的时变红外弱小目标检测前跟踪方法。 主要研究内容包括:(1)针对红外图像非平稳特性,将二维经验模态分解(EMD)方法引入红外图像预处理,拟研究基于偏微分方程的二维EMD杂波抑制方法;(2)为了降低时变多目标跟踪模型的状态维数、避免目标数过估,拟研究RBPF实现的多贝努利随机集滤波方法和基于对称相对熵的多贝努利项合并方法;(3)为减小检测判决延迟完成航迹确认,拟研究基于贝努利项多帧存在概率曲线的双统计量判决方法。 本项目研究成果将丰富目标数未知且时变的红外弱小目标检测前跟踪方法体系,项目研究将有效拓展和提高国防或民用领域对于低信噪比下弱小目标的检测能力,为红外探测系统性能改善提供理论支撑。
中文关键词: 弱小目标;随机集;检测前跟踪;多贝努利滤波;概率假设密度滤波
英文摘要: Infrared dim small targets detection in low signal-to-noise ratio is the hotspots and difficulties at domestic and overseas research areas, as low energy of signal, complex background and the number of targets time-varying. On the basis of random finite sets and combined with nonstationary signal processing methods the project will build time-varying multi-target track-before-detect system based on multi-Bernoulli random finite sets. This project will address the following aspects: (1)For the nonstationary feature of the infrared image , the two-dimensional empirical mode decomposition (EMD) is introduced to the clutter suppression and the two-dimensional empirical mode decomposition based on partial differential equation(PDE) will be discussed, which will obtain the candidate targets adaptively. (2) Multi-Bernoulli random finite sets filter based on Rao-Blackwellized particle filter and symmetric relative entropy will reduce the state dimension and computation, and improve the accuracy of target number estimation. (3)The present and absent dual statistics based on the multi-frame existence probability curve of Bernoulli item will reduce the declaration delays. The project is within the forefront of the discipline applied basic research. It will expand and improve the target identification and detection capabi
英文关键词: Dim small targets;Random finite set;Track before detect;Multi-Bernoulli filter;Probability hypothesis density filter