项目名称: 高光谱与极化SAR图像协同处理关键技术研究
项目编号: No.60872098
项目类型: 面上项目
立项/批准年度: 2009
项目学科: 金属学与金属工艺
项目作者: 张钧萍
作者单位: 哈尔滨工业大学
项目金额: 32万元
中文摘要: 高光谱(HSI)与极化合成孔径雷达(PolSAR)是近年来国内外遥感领域的研究热点。鉴于它们各自的优点与局限性,本项目在分别研究其数据特点、成像机理的基础上,重点研究互补信息的提取,探讨解决二者协同处理的关键技术,并利用仿真及实际数据对建立的模型和算法进行验证和评估。 本项目取得了以下主要成果:1)研究了HSI和PolSAR的成像原理与散射模型,进行了HSI的模型分析与复杂林地场景仿真、建立了PolSAR多成分散射模型(MCSM);2)提出了基于关键信息保持的相关向量机分类方法和基于支持向量数据描述的HSI解混与目标检测方法;3)提出了联合MCSM与支持向量机的城镇PolSAR图像分类及基于商空间粒度计算的PolSAR建筑物检测;4)针对不同信源图像时空分辨率不一致,研究了基于稳健特征提取与多特征/多约束的自适应多源图像高效配准技术;5)基于随机森林和特征与决策相结合的方法,研究了面向分类应用的HSI与PolSAR图像的协同处理技术。 项目的成果已公开发表于国内外著名杂志和重要国际会议上,得到同行的认可。将为相关领域的研究提供新思路和新方法,并推动我国高光谱与极化SAR的进一步应用。
中文关键词: 高光谱图像;极化SAR;协同处理;互补信息
英文摘要: Hyperspectral imaging (HSI) and Polarimetric SAR (PolSAR) are the hottest topics in remote sensing area in recent years. In view of their respective merits and limitations, this project focuses on complementary information extraction and association, and addresses the key technique issue of the synergic processing for HSI and PolSAR images, based on the analysis of data characteristic and imaging mechanism. Simulation and real data are used for experimental verification and evaluation of the models and algorithms proposed in the project. The outcomes of this project are as follows. 1) Research on the imaging principle and scattering model of HSI and PolSAR, analyze HSI and polarimetric HSI models and simulate a complex woodland scene, and set up multiple component scattering model (MCSM). 2) Investigate into key information preserving for HSI demand-oriented feature extraction, design and realize the classification based on relative vector machine as well as unmixing and target detection based on support vector data description. 3) Propose the classification method for PolSAR image combining MCSM with SVM and building detection based on granularity computing of quotient space theory using PolSAR image. 4) Research on robust feature extraction and multi-feature/multi-constraint adaptive registration techniques for multiple images. 5) Research and discussion on the classification-oriented synergic processing for HSI and PolSAR images on the basis of random forest and the fusion method combining the feature with decision. The achievements of the project have been published on international journals and conferences, which are accepted by other researchers in this area and which provide new methods and new ideas for the others. The outcomes of the research could play an important role to promote further applicaitons of HSI and PolSAR in our country.
英文关键词: Hyperspectral Image; Polarimetric SAR; Synergic Processing; Complementary Information