项目名称: 基于稳健统计的SAR图像配准方法研究
项目编号: No.61201323
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: 延伟东
作者单位: 西北工业大学
项目金额: 23万元
中文摘要: SAR图像配准是众多SAR图像应用不可或缺的组成部分,配准精度作为衡量图像配准方法的关键指标,对SAR图像后续处理具有重要影响。针对现有SAR图像配准算法稳健性较差、配准精度较低的研究现状,本项目围绕基于特征的SAR图像配准中两个核心问题- - 特征的构造和匹配模型的建立展开研究,探索建立SAR图像的稳健配准模型。特征构造方面,研究SAR图像散射中心的检测方法,并结合SAR图像的统计特性,构造散射中心的特征描述向量,获得具有不变性的SAR图像特征。匹配模型的建立方面,分析非同名特征对特征匹配精度的影响程度,在稳健统计理论框架内,提出非同名特征鉴别准则,在此基础上建立稳健特征匹配模型并设计快速匹配算法。本项目的完成将丰富和完善SAR图像配准的理论体系,为设计稳健的SAR图像配准方法提供理论依据。
中文关键词: 图像配准;稳健统计;影响函数;流形对齐;非同名特征
英文摘要: SAR image registration is the necessary technique in varies remote sensing applications.Registration accuracy is a key index for registration alogrithms, and has an important influence on downstream SAR image processing.Owing to neglect of SAR imaging properties, the robustness and accuracy of existing SAR image registration models are poor. This project centers around two fundamental issues of feature based SAR image registration: feature construction and establishment of feature matching model, to explore the roubst registration model of SAR image. For the feature construction, the project will study scattering centers extraction of SAR image, and construct robust local descriptors of scattering centers by combining with statistical properties of SAR image.For the establishment of feature matching model, the project will analyse the effect of each outlier on matching accuracy, and explore criteria to identify outliers in the framework of robust statistical theory.On this basis, the work here will also establish a robust feature matching model and propose a fast matching algorithm.This project will enrich and improve the theory of SAR image registration, and the fruits of the research will be exposed to a broader audience, thereby, helping establish new image registration models.
英文关键词: Image Registration;Robust Statistics;Influence Function;Manifold Alignment;Outliers