当前位置:科学网首页 > 基金首页 > 基于多目标最小优化理论的近景影像与三维激光点云几何配准方法研究

国家自然科学基金项目查询

基于多目标最小优化理论的近景影像与三维激光点云几何配准方法研究

批准号41601496 学科分类测量与地图学 ( D0108 )
项目负责人李彩林 负责人职称讲师 依托单位山东理工大学
资助金额19.00
万元
项目类别青年科学基金项目 研究期限2017 年 01 月 01 日 至
2019 年 12 月 31 日
中文主题词光学影像-激光雷达数据联合处理;点云配准与拼接;地面激光雷达;几何配准模型
英文主题词Joint processing of optical image and laser radar data;Point cloud registration and mosaicing;Terrestrial lidar;Geometric registration model

摘要

中文摘要 针对近景影像与三维激光点云几何配准方法存在精度不高、效率低下、模型不统一等问题,旨在建立近景影像与三维激光点云之间通用的几何配准数学模型,并研究精确稳健高效的几何配准算法。主要研究内容包括:(1)分析近景影像与三维激光点云数据之间内在的几何约束关系,基于多目标最优化理论,结合多类型观测值平差处理方法,建立多目标最小优化问题的几何配准数学模型;(2)设计能够消除镜头畸变引起的非刚性变形的几何配准算法,研究粗差剔除与最优化定权相结合的多目标最小优化模型精确稳定高效求解;(3)研究近景影像与三维激光点云几何配准质量评估方法,并在此基础上进一步深入探索,致力于发现其中的精度规律,为解决具体的工程实践问题提供思路和指导。本研究成果有望解决由于数据的异质性所带来的两种数据几何配准的难题,形成一套高精度稳定高效的近景影像与三维激光点云几何配准模型和方法,推动几何配准在数据融合等相关应用领域的进展。
英文摘要 For the inefficiency, low quality and no unified model of geometric registration method between close-range image and three-dimensional laser point cloud, the research project aims to establish general mathematical model of geometric registration between close-range image and three-dimensional laser point cloud, and study the accurate, robust and efficient geometric registration method. The main research contents include: (1) Study on intrinsic geometric constraints between close-range image and three-dimensional laser point cloud, combined with multi type observation adjustment method to establish geometric registration mathematical model of the multi-objective optimization problem based on the multi-objective optimization theory; (2) Design the geometric registration algorithm which can eliminate the non rigid deformation caused by lens distortion and study the method to solve the multi objective optimization model with high precision, stability and high efficiency by combining gross error elimination and the optimal method to determine the weight. (3) Study the quality assessment methods of geometry registration between close-range image and three-dimensional laser point cloud, and further explore the accuracy rules to provide ideas and guidance for solving practical problems in engineering practice. The results of this study ate expected to solve geometric registration problem of two kinds data from the heterogeneity. A stable, efficient and high precision geometrical registration model and methods should be established for the close range image and 3D laser point cloud, promoting the progress of geometric registration applications in the related fields of data fusion.
结题摘要

成果

序号 标题 类型 作者

关于我们| 网站声明| 服务条款| 联系方式| RSS| 中国科学报社 京ICP备14006957 京公网安备110402500057号
Copyright @ 2007- 中国科学报社 All Rights Reserved
地址:北京市海淀区中关村南一条乙三号   电话:010-62580783