[1]刘宇翔,张茂军,颜深,等.基于多任务学习的初始图像对选取方法[J].郑州大学学报(工学版),2021,42(01):56-62.[doi:10.13705/j.issn.1671-6833.2021.01.009]
　Liu Yuxiang,Zhang Maojun,Yan Shen,et al.Selecting Initial Image Pairs ba＜x＞sed on Multi-task Learning[J].Journal of Zhengzhou University (Engineering Science),2021,42(01):56-62.[doi:10.13705/j.issn.1671-6833.2021.01.009]
点击复制

基于多任务学习的初始图像对选取方法()
分享到：

《郑州大学学报(工学版)》[ISSN:1671-6833/CN:41-1339/T]

卷:
42
期数:
2021年01期
页码:
56-62
栏目:
出版日期:
2021-03-14

文章信息/Info

Title:
Selecting Initial Image Pairs ba＜x＞sed on Multi-task Learning
作者:
刘宇翔; 张茂军; 颜深; 李京蓓; 彭杨
国防科技大学系统工程学院;

Author(s):
Liu Yuxiang; Zhang Maojun; Yan Shen; Li Jingbei; Peng Yang;
School of System Engineering, University of Defense Science and Technology;

关键词:
Keywords:
DOI:
10.13705/j.issn.1671-6833.2021.01.009
文献标志码:
A
摘要:
初始图像对选取是增量式从运动中恢复结构的一个关键环节，但传统方法中存在计算效率低、对特殊场景不鲁棒的问题。因此，本文提出基于多任务学习的初始图像对选取网络以提高该过程的效率，并针对一些特殊场景提出结合场景连接图的选取策略，进一步提高重建的鲁棒性。最后，通过与传统SfM方法进行对比实验，证明了所提方法的有效性
Abstract:
The selection of the initial image pair is a key part of the incremental SfM, but the traditional method has the problems of low efficiency and not robust to special scenes. Therefore, this paper proposes an initial image pair selecting network ba＜x＞sed on multi-task learning to improve the efficiency of initial pair selecting, and a strategy combining scene connection graphs for some special scenes to improve the stability of reconstruction is added. Finally, the comparison experiment with the traditional SfM method proves the effectiveness of the proposed method.
更新日期/Last Update: 2021-03-15