低频超宽带合成孔径雷达叶簇隐蔽目标变化检测技术研究

项目名称： 低频超宽带合成孔径雷达叶簇隐蔽目标变化检测技术研究

项目编号： No.61302194

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： 王广学

作者单位： 中国人民解放军空军预警学院

项目金额： 24万元

中文摘要： 长期以来，低频UWB SAR叶簇隐蔽目标探测领域一直受到大量树干杂波虚警的困扰。为此，本项目拟开展低频UWB SAR叶簇隐蔽目标变化检测技术研究，以期利用树干杂波在不同时刻观测所得低频UWB SAR图像中具有很强相关性的特点，采用变化检测技术对其进行有效抑制。 本项目首先将对低频UWB SAR变化检测中的杂波建模方法进行研究，在此基础上提出相应的象素级变化检测方法；而后对低频UWB SAR图像特征信息提取技术进行研究，在此基础上提出相应的特征级变化检测法；最后开展基于像素级变化检测和特征级变化检测的融合变化检测方法研究。此外，本项目还将对变化检测中的图像配准预处理技术进行研究，并以不同场景的实测低频UWB SAR图像为基础，对本项目所提变化检测算法进行适应性分析。 本项目研究成果将有效改善低频UWB SAR叶簇隐蔽目标检测性能，对于提高我军战场监测能力具有重要意义。

中文关键词： 变化检测；超宽带合成孔径雷达；图像配准；统计分布特征；超球面支持向量机

英文摘要： Due to strong trunk clutter, foliage-concealed targets detection with low-frequency UWB SAR images has been troubled with high false alarm rate problem for a long period. For this reason, the technique of foliage-obscured target change detection based on low-frequency UWB SAR images is researched in this item. The object of the research is to effectively suppress trunk clutters by exploiting the strong correlation between trunk clutter in different temporal low-frequency UWB SAR images. In this project, we will firstly study the problem of clutter modeling in low-frequency UWB SAR change detection, and a pixel-based change detection algorithm will be proposed based on it. Then, the feature extraction method suitable for low-frequency UWB SAR images will be discussed, and a feature-based change detection algorithm will be designed based on it. At last, the fusion method for pixel-based change detection and feature-based change detection will be analyzed. Besides that, the image registration technique used in change detection will be also discussed in this project. And based on real low-frequency UWB SAR images of different scenes, we will test the adaptability of proposed change detection algorithms. The research of this project will effectively improve the performance of foliage obscured target detection based o

英文关键词： Change Detection；Ultra-Wide Band Synthetic Aperture Radar (UWB SAR)；Image Registration；Statistic Distribution Feature；HyperSphere Support Vector Machine (HS-SVM)