MH-Flow (Paper@VISAPP2026)
MH-Flow
Multi-View and Occlusion-Robust
Dense Correspondence Estimation
Based on Homographic Decomposition
VISAPP'2026
Thomas Vincent Chang1, Simon Seibt2, Kay Hartmann1, Bartosz von Rymon Lipinski1, Konrad Kluwak3 and Marc Erich Latoschik4
2 Institute for Production and Informatics, University of Applied Sciences Kempten, Germany
3 Department of Control and Quantum Computing, Faculty of Information and Communication Technology,
Wrocław University of Science and Technology, Poland
4 Human-Computer Interaction Group, University of Würzburg, Germany
Abstract
Dense correspondence estimation is an essential computer vision task, however, existing methods often struggle to produce accurate and reliable dense flow fields when encountering significant occlusions. This is primarily due to scenes with high depth complexity and wide baselines between input images. To address this challenge, MH-Flow is introduced, a novel approach specifically designed to generate occlusion-robust dense flow fields from two to four images. It leverages iterative dense feature matching to perform a multi-homography (MH) decomposition for an image pair. This decomposition allows to effectively detect and analyze occluded and non-occluded pixel regions. Furthermore, MH-Flow incorporates a multi-view strategy and matching transitivity to extrapolate flow information from two additional input views and to enhance robustness, particularly in challenging occluded areas. Experimental results across multiple datasets demonstrate MH-Flow’s significant superiority in handling large viewpoint changes and occlusions compared to leading CNN-based approaches, achieving more accurate and complete dense flow fields.
Paper
Pipeline Overview
Activity diagram of the dense correspondence pipeline.
MH-Flow first combines dense feature matching with edge-aware matching extrapolation and a Delaunay mesh to identify overlapping image regions. Homogeneous regions are transformed using their homographies, while inhomogeneous regions are subdivided and processed through segment correspondence search and validation. For non-overlapping or occluded regions, a third and fourth view can contribute transformation information through matching transitivity. Interpolation of the remaining gaps and variational refinement produce the final dense flow field.
Visual Results
BibTeX
@inproceedings{Chang2026MHFlow,
booktitle = {International Conference on Computer Vision Theory and Applications},
title = {{MH-Flow}: Multi-View and Occlusion-Robust Dense Correspondence Estimation Based on Homographic Decomposition},
author = {Chang, Thomas Vincent and Seibt, Simon and Hartmann, Kay and {von Rymon Lipinski}, Bartosz and Kluwak, Konrad and Latoschik, Marc Erich},
year = {2026},
publisher = {SCITEPRESS},
}
References
| [1] | Strecha, C., von Hansen, W., van Gool, L., Fua, P., Thoennessen, U.: On benchmarking camera calibration and multi-view stereo for high resolution imagery. In: IEEE Conference on Computer Vision and Pattern Recognition (2008). |
| [2] | Tan, D., Liu, J.-J., Chen, X., Chen, C., Zhang, R., Shen, Y., Ding, S., Ji, R.: ECO-TR: Efficient correspondences finding via coarse-to-fine refinement. In: European Conference on Computer Vision (2022). |
| [3] | Truong, P., Danelljan, M., Timofte, R., Van Gool, L.: PDC-Net+: Enhanced probabilistic dense correspondence network. IEEE Transactions on Pattern Analysis and Machine Intelligence, 45 (2023). |
| [4] | Truong, P., Danelljan, M., Timofte, R.: GLU-Net: Global-local universal network for dense flow and correspondences. In: IEEE Conference on Computer Vision and Pattern Recognition (2020). |
| [5] | Sun, D., Yang, X., Liu, M.-Y., Kautz, J.: PWC-Net: CNNs for optical flow using pyramid, warping, and cost volume. In: IEEE Conference on Computer Vision and Pattern Recognition (2018). |
| [6] | Chaurasia, G., Sorkine-Hornung, O., Drettakis, G.: Silhouette-aware warping for image-based rendering. Computer Graphics Forum, 30(4) (2011). |