项目名称: 结构张量与相位一致性联合约束的倾斜立体影像直线特征分级匹配
项目编号: No.41501492
项目类型: 青年科学基金项目
立项/批准年度: 2016
项目学科: 天文学、地球科学
项目作者: 陈敏
作者单位: 西南交通大学
项目金额: 20万元
中文摘要: 可靠的直线特征匹配是倾斜摄影测量数据处理面临的主要难题之一。由于倾斜立体影像之间存在较大程度的视角变化,并且直线特征邻域纹理信息较弱,导致传统的遥感影像直线特征匹配方法对倾斜立体影像难以获得可靠的匹配结果。为此,本项目拟利用结构张量利于表达影像局部区域几何信息以及相位一致性模型对影像对比度的不变性特性,研究一种可靠的倾斜立体影像直线特征匹配方法,主要研究内容包括:1)构造一种结构张量约束的特征区域定位方法,突破影像视角变化导致同名特征区域不一致的技术瓶颈;2)联合影像色彩信息,将相位一致性模型扩展到色彩空间,构建稳健的直线特征描述符,有效描述弱纹理区域信息;3)设计一套可靠、高效的分级匹配算法,有效提高匹配精度和效率。本项目研究有望解决倾斜立体影像直线特征匹配的难点问题,增强倾斜立体影像自动化处理能力,进而提高遥感影像处理和应用的水平。
中文关键词: 倾斜摄影测量;影像匹配;特征提取与定位;建筑物三维重建
英文摘要: Accurate and reliable straight line matching is one of the main challenges in oblique images processing and analysis. Traditional straight line matching methods can not obtain reliable matching results due to the significant viewpoint change between oblique images and the weak texture around straight line. In order to overcome the above problems, this project will introduce and utilize the advantages of structure tensor and phase congruency to research a reliable straight line matching method for oblique images. The main research contents of this project include: 1) construct a feature area localization method based on the constraint of structure tensor to break through the bottleneck of corresponding feature area computation under viewpoint change; 2) expand the original phase congruency model to color space and establish a robust straight line feature descriptor to describe weak texture information; 3) design a hierarchical matching algorithm to improve the matching accuracy and efficiency. This project is expected to solve the problems in oblique images straight line matching, enhance the automatic processing ability for oblique images, and then improve the performance of remote sensing image processing and application.
英文关键词: oblique photogrammetry;image matching;feature detection and localization;building 3D reconstruction