激光点云与数字影像融合的目标细部重建  

Fine reconstruction of objects by integrating close range point clouds and digital images

作　　者：林承达[1] 方益杭[1] 常婷婷[1] 季铮[2] 翟瑞芳[3] 

机构地区：[1]华中农业大学资源与环境学院,武汉430070 [2]武汉大学遥感信息工程学院,武汉430070 [3]华中农业大学理学院计算机系,武汉430070

出　　处：《计算机工程与应用》2015年第21期185-190,208,共7页Computer Engineering and Applications

基　　金：国家自然科学基金(No.41101409;No.41301522;No.41301518);高校博士学科点专项科研基金资助(No.20110146120012);教育部留学回国人员启动基金(No.4002-122010)

摘　　要：针对激光点云和高分辨率数字影像数据的优缺点,提出融合两种数据进行目标细部几何特征重建的方法。该方法以激光点云数据作为初始空间位置估计,就基于核线约束的多视影像匹配和基于物方的多视最小二乘影像匹配方法展开讨论,并以某小型文物为研究对象,探讨了集成近景激光扫描数据和高分辨率影像数据实现的目标细部几何特征重建的方法,通过实验验证了所提出方案的正确性和有效性。With the technological development in optics and electronics, range scanning systems are becoming more and more accurate, but more and more affordable. Since the range scanning systems directly capture the depth information of the world, they significantly simplify the analysis of range images. As a result, it is attracting more and more attention from both academia and industry. However, it can not capture the accurate information of break lines, which are usually expressed as semantic information in digital images. The characteristics of laser scanner data and camera data can be regarded as complementary, therefore, the integration of digital image and point clouds provides a promising approach for fine reconstruction of objects, especially catering to break liens and edge splits. This paper starts with the estimation proce-dures of the internal and external orientation parameters. The close range point clouds are then projected onto several digital images as the initial three dimensional coordinate values of some features, and then multi-view image matching theories are discussed in detail, which include matching based on epipolar lines, and matching based on least square adjustments. The performance of the developed procedures is evaluated through experimental results from real data. Experimental results show that the three dimensional points of some detailed features can be represented, which indicates that the proposed methodology is effective and efficient.

关 键 词：激光点云 数字影像 细部重建 多视匹配 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]