• 中国科学引文数据库(CSCD)来源期刊
  • 中文核心期刊中文科技核心期刊
  • Scopus RCCSE中国核心学术期刊
  • 美国EBSCO数据库 俄罗斯《文摘杂志》
  • 《日本科学技术振兴机构数据库(中国)》
二维码

隧道建设(中英文) ›› 2017, Vol. 37 ›› Issue (7): 808-815.DOI: 10.3973/j.issn.1672-741X.2017.07.005

• 研究与探索 • 上一篇    下一篇

基于ASIFT与RANSAC算法的岩体结构三维重建方法研究

熊自明1, 2, 董鑫1, 王明洋1, 2, 张清华1, 2, 马超1   

  1. (1. 解放军理工大学爆炸冲击防灾减灾国家重点实验室, 江苏南京 210007; 2. 南京理工大学机械工程学院, 江苏南京 210094)
  • 收稿日期:2017-01-03 修回日期:2017-06-15 出版日期:2017-07-20 发布日期:2017-07-26
  • 作者简介:熊自明(1980—),男,江西南昌人,2013年毕业于解放军信息工程大学,作战环境学专业,博士,讲师,现主要从事不良地质条件下工程安全风险评估方面的教学与研究工作。Email: xzm992311@163.com。
  • 基金资助:
    国家重点基础研究发展规划“973”项目(2013CB036005); 博士后基金(2017M611816); 国家自然科学基金青年项目(51608529)

Research on 3D Reconstruction Technology for Rock Mass Structure Based on ASIFT Algorithm and RANSAC Algorithm

XIONG Ziming1, 2, DONG Xin1, WANG Mingyang1, 2, ZHANG Qinghua1, 2, MA Chao1   

  1. (1. State Key Laboratory for Disaster Prevention & Mitigation of Explosion & Impact, PLA University of Science and Technology, Nanjing 210007, Jiangsu, China; 2. School of Mechanical Engineering, Nanjing University of Science and Technology, Nanjing 210094, Jiangsu, China)
  • Received:2017-01-03 Revised:2017-06-15 Online:2017-07-20 Published:2017-07-26

摘要:

隧道开挖过程中岩体结构面信息获取的快慢和准确程度直接影响隧道施工的安全,针对三维重建算法效率低的问题,提出一种基于ASIFT与RANSAC算法的岩体结构三维重建方法和策略。首先采用普通数码相机快速获取一系列真实描述岩体结构的数字影像; 然后针对岩体影像比例尺与倾斜角变化大、纹理相对贫乏的特点,采用ASIFT匹配算法与RANSAC误匹配点剔除算法来保证立体影像之间稀疏匹配的效率和可靠性; 最后采用光束法联合平差方法对系列影像之间的相对位置与姿态进行求解,并综合利用核线几何约束、半全局匹配、多视最小二乘匹配等策略实现岩体影像的密集匹配,最终提取出岩体的三维信息。工程实践表明: 基于ASIFT与RANSAC算法的岩体结构三维重建方法可以方便地实现工程岩体结构面完全非接触式测量,有效提高测量效率, 避免测量人员暴露于未支护岩体下的危险。

关键词: 岩体结构, 三维重建, 非接触测量, 掌子面影像, ASIFT算法, RANSAC算法

Abstract:

The acquisition and accuracy of information of rock mass discontinuities structural plane are the key to the safe tunnel excavation. A new 3D reconstruction method for rock mass structure based on ASIFT algorithm and RANSAC algorithm is proposed so as to improve the efficiency of 3D reconstruction algorithm. The ordinary digital camera is used to gain a series of digital images of real rock mass structure quickly. And then, according to the characteristics, i. e. large variation of rock mass image scale and angle of inclination, and poverty of texture of rock mass, the ASIFT matching algorithm and RANSAC mismatching point elimination algorithm are adopted to guarantee the efficiency and reliability of the sparse matching between stereo images. Finally, the bundle adjustment method in conjunction with adjustment method is used to solve the relative position and attitude between image sequences. Meanwhile, the methods of epipolar line geometric constraint, localfull matching and least square matching are comprehensively adopted to achieve the intensive matching of rock mass images; and then 3D information of rock mass structure is finally extracted. The engineering practice shows that the noncontact measurement of rock mass structural planes can be completed conveniently by the abovementioned method; meanwhile, the efficiency can be improved and the risk of survey crew exposing under rock mass can be avoided.

Key words: rock mass structure, 3D reconstruction, noncontact measurement, image of tunnel face, ASIFT algorithm, RANSAC algorithm

中图分类号: