基于压缩感知的大尺度SAR图像三维重建及其算法的研究

项目名称： 基于压缩感知的大尺度SAR图像三维重建及其算法的研究

项目编号： No.61472278

项目类型： 面上项目

立项/批准年度： 2015

项目学科： 计算机科学学科

项目作者： 温显斌

作者单位： 天津理工大学

项目金额： 82万元

中文摘要： 合成孔径雷达（SAR）图像城区目标的三维重建在城市规划、监测等领域蕴藏着广泛的应用前景，然而由于SAR特殊的成像机理，对SAR图像的理解造成极大困难，制约着SAR技术的推广应用。本项目针对SAR图像的特点，以城市应用为突破口，从研究城区典型地物目标的散射性质与机理入手，研究城区复杂环境下地物目标对SAR信号的散射特性，探讨SAR对城区成像机理的认识，在此基础上，融合虚拟现实技术，开展基于压缩感知和图模型的SAR图像目标快速三维重建模型与方法技术的研究，力图突破基于SAR图像的快速三维重建关键技术，包括SAR图像中目标检测、特征提取、基于特征与轮廓的图像配准、基于压缩感知的大尺度SAR图像三维模型及其优化算法的构建、理论分析与实现等，为今后基于SAR图像的三维可视化分析打下坚实基础。

中文关键词： 三维重建；压缩感知；图像配准；计算机视觉；图像分析

英文摘要： The three-dimensional reconstruction of objects from Synthetic Aperture Radar (SAR) images is widely applied in the field of urban planning and monitoring. However, the understanding of the SAR images is very difficult due to the special imaging mechanism of SAR, which restricts the promotion and application of SAR technology.This project, using the characteristics of SAR images, taking the application of City as a breakthrough, starting from the study of scattering nature and mechanism of the surface features, researchs scattering features of the target on the SAR signal in the complex environment. Meanwhile, the project explores the tecnhologies of ulitizing the imaging mechanism of SAR on the city.On this basis, combining the virtual reality technology, the fast three-dimensional reconstruction model, methods and techniques of urban target are developed based on compressed sensing and graph model from SAR images. Trying to make breakthroughs in fast key technologies of the three-dimensional reconstruction based on SAR image, including target detection,feature extraction, image registration based on features and contour, three-dimensional model and its optimization algorithms for large-scale SAR image based on compressed sensing. All of these make a good foundation for the studing about three-dimensional visual analysis on SAR images.

英文关键词： 3D reconstruction;Compressed sensing;Imagegistration re;Computure vision;Image analysis