Long-term Structural Displacement Monitoring using Image Sequences and Spatio-Temporal Context Learning
Long-term Structural Displacement Monitoring using Image Sequences and Spatio-Temporal Context Learning
复制标题
使用图像序列和时空上下文学习进行长期结构位移监测
DOI：
--
复制
发表时间：
2017
期刊：
影响因子：
--
通讯作者：
F. Catbas
中科院分区：
文献类型：
--
作者：
C. Dong;O. Celik;F. Catbas
在这项研究中，基于视觉的位移测量方法，使用图像序列和时空上下文（STC）学习介绍了长期的结构位移监测。通过与现有的视觉位移测量方法（DIC、FLANN-SURF和LK-SURF）以及LVDT在不同恶劣测量条件下（包括光照变化和人工雾引起的随机遮挡）的地面真实值的对比研究，验证了该方法的可行性。实验结果表明，与现有的基于视觉的方法相比，该方法对光照变化和随机遮挡具有更好的鲁棒性。该方法在处理长期的现场结构位移监测任务是有前途的。
In this study, a vision-based displacement measurement method using image sequences and spatio-temporal context (STC) learning is introduced for long-term structural displacement monitoring. Comparative study is carried out to verify the feasibility of the proposed method with current vision-based displacement measurement methods including (DIC, FLANN-SURF and LK-SURF) and the ground truth from LVDT under different adverse measuring conditions (including illumination changes and random occlusion induced by artificial mist). The results show that the proposed method has better robustness to illumination changes and random occlusion than current vision-based methods. The proposed method is promising in handling long-term structural displacement monitoring task in field.