To address the information loss caused by viewpoint changes in cross-view image geo-localization, which degrades localization accuracy, a cross-view geo-localization method for limited field-of-view images with fused azimuthal information (FAECGeo) is proposed. Firstly, a feature modulation module with fused azimuthal information is designed to enhance image feature maps from the corresponding image locations, while predicting an azimuthal information confidence map to adaptively fuse the enhanced features with the original ones, so as to optimize the utilization of azimuthal information and semantic information. Then, an LogSumExp (LSE) pooling module weighted by azimuthal information confidence is introduced to reweight the accumulated exponentials, reducing interference caused by overemphasized azimuthal information in global representations and improving feature robustness. Experimental results show that the localization accuracy of FAECGeo outperforms existing state-of-the-art methods under different field-of-view settings. When the field-of-view is 90°, the maximum R@1 scores on the CVUSA and CVACT datasets are improved by 3.7 and 3.2 percentage points, respectively.
GEPPERTM, LARSSONV, SPECIALEP, et al. Privacy preserving localization and mapping from uncalibrated cameras[C]//Proceedings of the 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway, USA: IEEE, 2021:1809-1819.
[5]
HUS X, FENGM D, NGUYENR M H, et al. CVM-net: cross-view matching network for image-based ground-to-aerial geo-localization[C]//Proceedings of the 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway, USA: IEEE, 2018:7258-7267.
[6]
RADFORDA, KIMJ W, HALLACYC, et al. Learning transferable visual models from natural language supervision[C]//Proceedings of the 38th International Conference on Machine Learning. New York, USA: PMLR, 2021: 8748-8763.
SHIY J, YUX, CAMPBELLD, et al. Where am I looking at? joint location and orientation estimation by cross-view matching[C]//Proceedings of the 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway, USA: IEEE, 2020:4063-4071.
[9]
ZHUS J, SHAHM, CHENC. TransGeo: transformer is all you need for cross-view image geo-localization[C]//Proceedings of the 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway, USA: IEEE, 2022:1152-1161.
[10]
SHUGAEVM, SEMENOVI, ASHLEYK, et al. ArcGeo: localizing limited field-of-view images using cross-view matching[C]//Proceedings of the 2024 IEEE/CVF Winter Conference on Applications of Computer Vision. Piscataway, USA: IEEE, 2024:208-217.
[11]
WANGW, CUIJ K, CAOX W, et al. Vaclnet: view augmentation contrastive learning for limited FOV cross-view geo-localization[C]//Proceedings of the 2025 IEEE International Geoscience and Remote Sensing Symposium. Piscataway, USA: IEEE, 2025:2177-2181.
[12]
RODRIGUESR, TANIM. Global assists local: effective aerial representations for field of view constrained image geo-localization[C]//Proceedings of the 2022 IEEE/CVF Winter Conference on Applications of Computer Vision. Piscataway, USA: IEEE, 2022:2694-2702.
[13]
LIC R, YANC, XIANGX J, et al. AMPLE: automatic progressive learning for orientation unknown ground-to-aerial geo-localization[J]. IEEE Transactions on Geoscience and Remote Sensing, 2025,63:No.5800115.
[14]
HUB N, LIH W, BUS H, et al. OriLoc: unlimited-FoV and orientation-free cross-view geolocalization[J]. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2025,18:15508-15522.
[15]
MIL, XUC, CASTILLO-NAVARROJ, et al. ConGeo: robust cross-view geo-localization across ground view variations[C]//Proceedings of the 18th European Conference on Computer Vision. Cham, Switzerland: Springer, 2024:214-230.
[16]
CAIK Y, LIX, TIANF H, et al. GeoMatch: multi-view contrastive learning for limited CVGL with semantic uncertainty[C]//Proceedings of the 3rd International Workshop on UAVs in Multimedia: Capturing the World from a New Perspective. New York, USA: ACM,2025: 10-14.
[17]
ZHANGX H, LIX Y, SULTANIW, et al. Cross-view geo-localization via learning disentangled geometric layout correspondence[C]//Proceedings of the 37th AAAI Conference on Artificial Intelligence. Palo Alto, USA: AAAI Press, 2023,37(3):3480-3488.
[18]
ZHUY Y, YANGH J, LUY X, et al. Simple, effective and general: a new backbone for cross-view image geo-localization[DB/OL]. (2023-02-03)[2026-02-09].
[19]
DEUSERF, HABELK, OSWALDN. Sample4Geo: hard negative sampling for cross-view geo-localisation[C]//Proceedings of the 2023 IEEE/CVF International Conference on Computer Vision. Piscataway, USA: IEEE, 2023:16801-16810.