基于杂波统计建模的SAR图像目标智能快速检测方法研究

项目名称： 基于杂波统计建模的SAR图像目标智能快速检测方法研究

项目编号： No.40801179

项目类型： 青年科学基金项目

立项/批准年度： 2009

项目学科： 金属学与金属工艺

项目作者： 高贵

作者单位： 中国人民解放军国防科学技术大学

项目金额： 19万元

中文摘要： 从复杂的大幅SAR场景图像中有效提取目标区域的检测技术是当前对地观测的基础前沿课题，具有重要的学术价值和应用价值。为发展快速智能的SAR图像目标自动检测技术，项目所提方案主要分两步展开：一是在综合考虑目标检测算法的精度和实时性要求，深入研究现有杂波统计模型的精度、参数估计的计算复杂度，并在理论分析杂波统计模型相互关系的基础上研究了SAR图像杂波统计建模技术，核心是最优杂波统计模型库的建立；二是在最优杂波统计模型库建立的基础上，构建快速智能的CFAR算法，并根据滑窗对图像遍历检测过程中窗口内杂波数据的"分裂－合并"特点，通过引入杂波统计量的迭代计算方式，实现目标的CFAR快速检测。项目研究成果将推动并丰富SAR图像目标检测理论和技术研究。

中文关键词： 合成孔径雷达（SAR）；目标；杂波；检测；统计建模

英文摘要： Aiming at effectively extracting ground vechile targets, the techniques of extracting target regions from SAR images with complex scene content are among the basic and frontal subjects of the applications of earth observation. This project wants to develop a new scheme of target detection in SAR images, which is fast and intelligent. The proposed scheme consists of two steps: firstly, under the consideration of the requests of accuracy and practicalility for target detection algorithm, we study comprehensively the accuracy of existing statistical models of clutter, the complex degree of parameter estimation. Moreover, based on analyzing the relationship of the existing models in theory, we study statistical modeling of clutter in SAR images, whose key problem is the construction of the optimal statistical model sets of clutter; Secondly, on the basis of the optimal statistical model sets, the CFAR algorithm that is fast, intelligent is designed. Then, according to the splitting-merging characteristic of data during the detection process of prescreening images by sliding-window, the fast CFAR target detection is obtained by introducing the iteration computation of the clutter statistics. The theory and tecnique studies of target detection in SAR images will be furtherly enriched and promoted.

英文关键词： Synthetic Aperture Radar; Target; Clutter; Detection; Statistical Modeling