TomoFluid: Reconstructing Dynamic Fluid From Sparse View Videos

Abstract
Visible light tomography is a promising and increasingly popular technique for fluid imaging. However, the use of a sparse number of viewpoints in the capturing setups makes the reconstruction of fluid flows very challenging. In this paper, we present a state-of-the-art 4D tomographic reconstruction framework that integrates several regularizers into a multi-scale matrix free optimization algorithm. In addition to existing regularizers, we propose two new regularizers for improved results: a regularizer based on view interpolation of projected images and a regularizer to encourage reprojection consistency. We demonstrate our method with extensive experiments on both simulated and real data.

Citation
Zang, G., Idoughi, R., Wang, C., Bennett, A., Du, J., Skeen, S., … Heidrich, W. (2020). TomoFluid: Reconstructing Dynamic Fluid From Sparse View Videos. 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). doi:10.1109/cvpr42600.2020.00194

Acknowledgements
This work was supported by KAUST as part of VCC and CCRC Center Competitive Funding and KAUST Competitive Research Grants. We thank the anonymous reviewers for their insightful comments, and Yuansi Tian for helping with the data collection.

Publisher
Institute of Electrical and Electronics Engineers (IEEE)

Conference/Event Name
2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)

DOI
10.1109/CVPR42600.2020.00194

Additional Links
https://ieeexplore.ieee.org/document/9156450/https://ieeexplore.ieee.org/document/9156450/https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=9156450

Permanent link to this record