基于几何-电磁建模的高分辨率SAR图像建筑物高度信息提取技术

项目名称： 基于几何-电磁建模的高分辨率SAR图像建筑物高度信息提取技术

项目编号： No.61201338

项目类型： 青年科学基金项目

立项/批准年度： 2013

项目学科： 电子学与信息系统

项目作者： 赵凌君

作者单位： 中国人民解放军国防科学技术大学

项目金额： 23万元

中文摘要： 本项目面向高分辨率SAR图像在城区环境遥感中的应用需求，旨在充分挖掘建筑物典型影像特征与其几何结构、电磁散射特性之间的联系，探索可用于高度提取的新特征，实现准确性高、灵活性好、可扩展性强的建筑物高度信息提取。首先构建几何-电磁散射模型，全面揭示建筑物SAR成像机理，定量描述影像特征与建筑物高度之间的关系；在此基础上，提出基于特征预测-匹配的建筑物高度信息提取新思路，可综合利用几何、散射等多种特征；并在特征匹配中引入对特征不确定性的定量表征，提高高度信息提取的可靠性。项目研究成果将为高分辨率SAR图像建筑物解译提供新的技术途径，对充分挖掘SAR图像信息、提高对高分辨率SAR图像中复杂目标的解译能力也具有重要的参考价值。

中文关键词： SAR；建筑物提取；高度估计；不确定特征；

英文摘要： With the increasing demand on high-resolution SAR image interpretation of urban areas, this project aims at analyzing the relationship between building features in SAR images and its geometrical structure and electromagnetic properties, exploring potential features for retrieving building height, and developing height retrieval method with high accuracy, flexibility and extensibility. By constructing geometrical-electromagnetic model, the SAR imaging mechanism of buildings is illustrated, and the relationship between image features and building height is quantitatively described. Based on the model, a novel way of building height retrieval based on feature predicting-matching is proposed, which can utilizes multi-features such as geometrical and scattering features. With the introduction of feature uncertainty modeling in feature matching, the performance of building height estimation can be improved. Our research will provide a new way for building information extraction from high-resolution SAR images. It is also valuable for fully exploiting the large amount of information present in SAR images and improving the interpretation capability of complex objects.

英文关键词： SAR；building extraction；height estimation；uncertain feature；