Rights statement: This is the author’s version of a work that was accepted for publication in Pattern Recognition. Changes resulting from the publishing process, such as peer review, editing, corrections, structural formatting, and other quality control mechanisms may not be reflected in this document. Changes may have been made to this work since it was submitted for publication. A definitive version was subsequently published in Pattern Recognition, 138, 2023 DOI: 10.1016/j.patcog.2023.109442
Accepted author manuscript, 6.5 MB, PDF document
Available under license: CC BY-NC-ND: Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License
Final published version
Research output: Contribution to Journal/Magazine › Journal article › peer-review
Research output: Contribution to Journal/Magazine › Journal article › peer-review
}
TY - JOUR
T1 - Efficient large-scale oblique image matching based on cascade hashing and match data scheduling
AU - Zhang, Qiyuan
AU - Zheng, Shunyi
AU - Zhang, Ce
AU - Wang, Xiqi
AU - Li, Rui
N1 - This is the author’s version of a work that was accepted for publication in Pattern Recognition. Changes resulting from the publishing process, such as peer review, editing, corrections, structural formatting, and other quality control mechanisms may not be reflected in this document. Changes may have been made to this work since it was submitted for publication. A definitive version was subsequently published in Pattern Recognition, 138, 2023 DOI: 10.1016/j.patcog.2023.109442
PY - 2023/6/30
Y1 - 2023/6/30
N2 - In this paper, we design an efficient large-scale oblique image matching method. First, to reduce the number of redundant transmissions of match data, we propose a novel three-level buffer data scheduling (TLBDS) algorithm that considers the adjacency between images for match data scheduling from disk to graphics memory. Second, we adopt the epipolar constraint to filter the initial candidate points of cascade hashing matching, thereby significantly increasing the robustness of matching feature points. Comprehensive experiments are conducted on three oblique image datasets to test the efficiency and effectiveness of the proposed method. The experimental results show that our method can complete a match pair within 2.50∼2.64 ms, which not only is much faster than two open benchmark pipelines (i.e., OpenMVG and COLMAP) by 20.4∼97.0 times but also have higher efficiency than two state-of-the-art commercial software (i.e., Agisoft Metashape and Pix4Dmapper) by 10.4∼50.0 times.
AB - In this paper, we design an efficient large-scale oblique image matching method. First, to reduce the number of redundant transmissions of match data, we propose a novel three-level buffer data scheduling (TLBDS) algorithm that considers the adjacency between images for match data scheduling from disk to graphics memory. Second, we adopt the epipolar constraint to filter the initial candidate points of cascade hashing matching, thereby significantly increasing the robustness of matching feature points. Comprehensive experiments are conducted on three oblique image datasets to test the efficiency and effectiveness of the proposed method. The experimental results show that our method can complete a match pair within 2.50∼2.64 ms, which not only is much faster than two open benchmark pipelines (i.e., OpenMVG and COLMAP) by 20.4∼97.0 times but also have higher efficiency than two state-of-the-art commercial software (i.e., Agisoft Metashape and Pix4Dmapper) by 10.4∼50.0 times.
KW - Cascade hashing
KW - Feature point matching
KW - Match data scheduling
KW - Oblique image matching
KW - SIFT
KW - Structure from motion
U2 - 10.1016/j.patcog.2023.109442
DO - 10.1016/j.patcog.2023.109442
M3 - Journal article
VL - 138
JO - Pattern Recognition
JF - Pattern Recognition
SN - 0031-3203
M1 - 109442
ER -