The problem of dim target detection in sea clutter, which faces the challenge of low SCR and high false alarm, is studied in this project. This project includes 3 main research areas, the continuous multi-pulse coherent sea clutter modeling, the track-before-detect applying the characteristics of sea clutter to the Random Finite Set (RFS), and long time coherent integration under the constraint of target trajectory. Based on the compound Gaussian model, the sea clutter simulator meeting the needs of multi-frame integration research will be proposed, which includes the process of sea spike rearrangement, spectral shaping filtering and pulse-pulse correlation. The proposed track-before-detect method will be more consistent with the sea scene because the inverse Gamma distribution and extracted features are applied to the RFS filtering. According to the obtained target track, the Doppler frequency can be extracted by the spectral similarity between two neighboring range bins. The ROI of the target can be extracted on the RD domain, and then the combined parameter searching for long time coherent integration is transformed into phase compensation and coherent integration in the ROI regions. Through the research of this project, some new ideas and methods will be proposed for multi-frame integration of the dim target detection in sea clutter. Moreover, it also provides reference for dim target detection in other complex backgrounds.
本项目针对海杂波背景下弱目标检测所面临的低信杂比、高虚警等问题,研究连续多脉冲相参海杂波建模方法、海杂波特性与随机有限集(RFS)框架相结合的检测前跟踪方法和航迹约束下的长时间相参积累方法。在复合高斯模型的基础上,通过海尖峰重排、频谱整形滤波、脉间相关等处理,给出满足多帧积累算法研究需求的海杂波仿真模型;将海杂波纹理分量服从的逆伽马分布引入RFS滤波,并将提取的海杂波特征应用于滤波过程中的自适应参数调整,给出与海面场景相吻合的检测前跟踪方法;根据所获取的目标航迹,通过邻近距离门的频谱相似性比较提取运动目标的多普勒频率范围,在距离-多普勒域上提取目标的ROI,将长时间相参积累的组合参数搜索转化为ROI区域内的相位补偿和相参积累。通过本项目研究,为海杂波背景下基于多帧积累的弱目标检测提供新的思路和方法,同时也为其他复杂背景下的弱目标检测研究提供借鉴。
本项目针对海杂波背景下弱小目标的检测问题,开展了相参多脉冲海杂波建模、基于随机有限集的海上目标检测、基于长时间相参积累的海上目标检测、基于激光雷达的海上小目标检测前跟踪等技术研究。给出了能够反映海尖峰、纹理、脉间相关性、多普勒谱特性的海杂波仿真模型,提出了强杂波环境下基于随机集滤波器的弱目标检测方法,提出了基于ROI提取和分阶段补偿的海上目标长时间相参积累方法,并搭建了基于激光雷达的海上近距离目标检测系统。南非实测数据的实验结果表明了所建立的仿真模型和所提出的目标检测方法的有效性,另外,实测激光点云也初步验证了激光雷达在海上弱小目标检测方面的有效性。
{{i.achievement_title}}
数据更新时间:2023-05-31
基于全模式全聚焦方法的裂纹超声成像定量检测
基于图卷积网络的归纳式微博谣言检测新方法
一种改进的多目标正余弦优化算法
人工智能技术在矿工不安全行为识别中的融合应用
面向工件表面缺陷的无监督域适应方法
海杂波背景下漂浮目标和低空目标检测方法研究
海杂波背景下微弱信号特征提取及检测方法研究
海杂波中弱目标微动特征提取及检测技术研究
复杂背景下弱目标联合检测与跟踪理论和方法研究