地面激光扫描点云和光学图像的球面二次成像模型及多球面组合自动配准研究

项目名称： 地面激光扫描点云和光学图像的球面二次成像模型及多球面组合自动配准研究

项目编号： No.41201484

项目类型： 青年科学基金项目

立项/批准年度： 2013

项目学科： 地理学

项目作者： 张毅

作者单位： 武汉大学

项目金额： 24万元

中文摘要： 为融合地面激光扫描点云和光学图像的空间信息，针对两类数据在成像方式、反射特征、空间基准等方面的差异性，根据点云和图像的成像模式和球面成像原理，研究球面二次成像模型,建立点云和图像的球面模型参数联合标定方法。研究点云激光反射强度信息的辐射校正方法，解决同质目标由于空间形态和表面状态所产生的图像异质问题，实现球面图像的灰度改正和特征恢复。研究球面图像上的特征提取、伪特征消除和特征匹配方法。针对点云和图像的多种组合模式，深入研究单一组合模式和复杂组合模式下的配准模型建立和解算，形成一套基于球面系统的二次成像模型及多球面组合配准方法，实现大范围、多测站、多视角地面激光扫描点云和光学图像的快速自动配准。

中文关键词： 地面激光扫描；球面成像；图像匹配；球面核线；图像配准

英文摘要： For integrating spatial information of terrestrial laser scanning point cloud and optical image, this project is aimed at two categories of data in imaging modalities, reflection characteristics, spatial reference and other aspects of the differences, using the spherical geometry system to solve the problem of automatic registration. Based on the point cloud and the imaging model and spherical imaging principle, we research on spherical imaging model and establish the spherical model parameter calibration method. In order to solve the heterogeneity problems produced by homogeneous target due to the spatial morphology and surface state of the image, we research on radiation correction method of laser reflection intensity, and achieve the spherical image gray correction and feature restoration. Then we research on feature extraction, pseudo feature elimination and feature matching method on spherical image. For multiple combination modes of point clouds and images, we research on registration model and solution in single combination pattern and complex combination pattern in depth. A set of imaging model based on spherical system and multiple spherical combined registration method can be formed, so as to achieve fast automatic registration of terrestrial laser scanning point cloud and optical image in large scale,

英文关键词： Terrestrial Laser Scanning；Spherical Imaging；Image Matching；Spherical Epipolar Line；Image Registration