测绘通报 ›› 2022, Vol. 0 ›› Issue (8): 81-86.doi: 10.13474/j.cnki.11-2246.2022.0236

• 学术研究 • 上一篇    下一篇

基于相容性任务分配的三维建模集群方法

吴彦格1, 葛英伟2, 许国政3, 郭丙轩2, 彭哲3   

  1. 1. 中国中铁二院工程集团有限公司, 四川 成都 610031;
    2. 武汉大学测绘遥感信息工程国家重点实验室, 湖北 武汉 430079;
    3. 武汉讯图科技有限公司, 湖北 武汉 430079
  • 收稿日期:2022-04-15 发布日期:2022-09-01
  • 通讯作者: 葛英伟。E-mail:yingwei.g@whu.edu.cn
  • 作者简介:吴彦格(1974-),女,高级工程师,主要从事航空摄影与遥感专业相关工作。E-mail:1041607489@qq.com
  • 基金资助:
    国家自然科学基金重大研究计划(B2021061516)

A 3D modeling cluster approach based on compatible task assignment

WU Yange1, GE Yingwei2, XU Guozheng3, GUO Bingxuan2, PENG Zhe3   

  1. 1. China Railway Eryuan Engineering Group Co., Ltd., Chengdu 610031, China;
    2. State Key Laboratory ofInformation Engineering in Surveying, Mapping & Remote Sensing, Wuhan University, Wuhan 430079, China;
    3. Wuhan Xuntu Technology Co., Ltd., Wuhan 430079, China
  • Received:2022-04-15 Published:2022-09-01

摘要: 针对现有集群方法在倾斜影像三维建模时节点上存在的CPU和GPU资源浪费现象及三维建模效率不高等问题,本文提出了基于相容性任务分配的集群三维建模方法。首先,通过对三维重建任务进行两步提交划分,在建模效果不变的前提下提升集群的执行效率。然后,针对集群节点资源浪费现象,提出了相容性任务节点分配方式,实现了基于相容性任务分配的三维建模集群。最后,进行大规模影像三维重建试验测试,验证了大数据试验环境下相容性理论对分配方式优化的优越性,证明了本文方法对三维建模效率有明显提升作用。

关键词: 三维重建, 实景建模, 集群系统, 相容性算法, 倾斜影像

Abstract: In response to the existing clustering methods that have wasted CPU and GPU resources on nodes during 3D modeling of tilt images and the inefficiency of 3D modeling,this paper proposes a cluster 3D modeling method based on compatibility. Firstly,by dividing the 3D reconstruction task based on two-step submission,the execution efficiency of the cluster is improved while the modeling effect remains unchanged. Then,we propose the compatibility task node allocation method for the waste of cluster node resources,and complete the implementation of the 3D modeling cluster method based on the compatibility task allocation. Finally,the experimental test of large-scale image 3D reconstruction is conducted to verify the superiority of the compatibility theory on the optimization of the allocation method under the experimental environment of big data,which proves that the algorithm of this paper has significantly improved the efficiency of 3D modeling.

Key words: 3D reconstruction, real-world modeling, cluster systems, compatibility algorithms, tilt image

中图分类号: