EDM: Equirectangular Projection-Oriented Dense Kernelized Feature Matching
Fuente:
arXiv
Enregistré dans:
| Auteurs principaux: | , , , , , , |
|---|---|
| Format: | Preprint |
| Publié: |
2025
|
| Sujets: | |
| Accès en ligne: | |
| Tags: |
Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
|
| _version_ | 1866929735341178880 |
|---|---|
| author | Jung, Dongki Choi, Jaehoon Lee, Yonghan Jeong, Somi Lee, Taejae Manocha, Dinesh Yeon, Suyong |
| author_facet | Jung, Dongki Choi, Jaehoon Lee, Yonghan Jeong, Somi Lee, Taejae Manocha, Dinesh Yeon, Suyong |
| contents | We introduce the first learning-based dense matching algorithm, termed Equirectangular Projection-Oriented Dense Kernelized Feature Matching (EDM), specifically designed for omnidirectional images. Equirectangular projection (ERP) images, with their large fields of view, are particularly suited for dense matching techniques that aim to establish comprehensive correspondences across images. However, ERP images are subject to significant distortions, which we address by leveraging the spherical camera model and geodesic flow refinement in the dense matching method. To further mitigate these distortions, we propose spherical positional embeddings based on 3D Cartesian coordinates of the feature grid. Additionally, our method incorporates bidirectional transformations between spherical and Cartesian coordinate systems during refinement, utilizing a unit sphere to improve matching performance. We demonstrate that our proposed method achieves notable performance enhancements, with improvements of +26.72 and +42.62 in AUC@5° on the Matterport3D and Stanford2D3D datasets. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2502_20685 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | EDM: Equirectangular Projection-Oriented Dense Kernelized Feature Matching Jung, Dongki Choi, Jaehoon Lee, Yonghan Jeong, Somi Lee, Taejae Manocha, Dinesh Yeon, Suyong Computer Vision and Pattern Recognition We introduce the first learning-based dense matching algorithm, termed Equirectangular Projection-Oriented Dense Kernelized Feature Matching (EDM), specifically designed for omnidirectional images. Equirectangular projection (ERP) images, with their large fields of view, are particularly suited for dense matching techniques that aim to establish comprehensive correspondences across images. However, ERP images are subject to significant distortions, which we address by leveraging the spherical camera model and geodesic flow refinement in the dense matching method. To further mitigate these distortions, we propose spherical positional embeddings based on 3D Cartesian coordinates of the feature grid. Additionally, our method incorporates bidirectional transformations between spherical and Cartesian coordinate systems during refinement, utilizing a unit sphere to improve matching performance. We demonstrate that our proposed method achieves notable performance enhancements, with improvements of +26.72 and +42.62 in AUC@5° on the Matterport3D and Stanford2D3D datasets. |
| title | EDM: Equirectangular Projection-Oriented Dense Kernelized Feature Matching |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2502.20685 |