基于排名聚合的跨视角图像地理定位方向鲁棒性增强方法*

doi:10.7523/j.ucas.2025.019

中国科学院大学学报

• •

基于排名聚合的跨视角图像地理定位方向鲁棒性增强方法^*

盛怡宁^1,2, 赵理君^1†, 张正¹, 唐娉¹

1 中国科学院空天信息创新研究院,北京, 100094;
2 中国科学院大学电子电气与通信工程学院,北京, 100049

收稿日期:2025-01-17 修回日期:2025-04-07
通讯作者: ^†E-mail: zhaolj201934@aircas.ac.cn
基金资助:
^*“十四五”民用航天技术预先研究项目（D040404）、中国科学院空天信息创新研究院“未来之星”人才计划（2021KTYWLZX07）、中国科学院青年创新促进会（2022127）资助

A ranking aggregation-based method for enhancing directional robustness in cross-view image geo-localization

SHENG Yining^1,2, ZHAO Lijun¹, ZHANG Zheng¹, TANG Ping¹

1 Aerospace Information Research Institute, Chinese Academy of Sciences, Beijing 100094, China;
2 School of Electronic, Electrical and Communication Engineering, University of Chinese Academy of Sciences, Beijing 100049, China

Received:2025-01-17 Revised:2025-04-07

摘要/Abstract

摘要： 跨视角图像地理定位（cross-view image geo-localization,CVGL）是一项通过多源图像实现地理位置匹配的关键技术,广泛应用于机器人导航、自动驾驶等领域。然而,现有方法多基于影像方向对齐的假设,难以适应实际场景中的方向偏差问题。为此,本文提出了一种基于排名聚合的方向鲁棒性增强方法,通过对地面影像进行随机方向偏移,结合均值聚合和最小值聚合策略提升匹配准确性与鲁棒性。在CVUSA_360和CVACT_360数据集上的实验表明,该方法显著提升了不同模型在方向未对齐场景下的性能,尤其在R@1指标上表现突出。可视化分析进一步验证了其在增强方向鲁棒性和抑制噪声方面的优势,为复杂场景下的地理定位任务提供了高效的解决方案。

关键词: 跨视角图像地理定位, 方向未对齐, 排名聚合, 方向鲁棒性

Abstract: Cross-view image geo-localization (CVGL) is a key technique for achieving geographic location matching through multi-source images, with broad applications in fields such as robot navigation and autonomous driving. However, existing methods typically assume that images are directionally aligned, making them less effective in real-world scenarios with directional deviations. To address this challenge, this paper proposes a ranking aggregation-based method for enhancing directional robustness. By introducing random directional shifts to ground-level images and employing mean and min aggregation strategies, the proposed method improves the matching accuracy and robustness. Experiments conducted on the CVUSA_360 and CVACT_360 datasets demonstrate that this approach significantly enhances the performance of various models under directionally misaligned conditions, with particularly outstanding results in terms of the R@1 metric. Visual analysis further verifies the advantages of the method in strengthening directional robustness, suppressing noise, and improving matching accuracy, thus providing an efficient solution for geo-localization tasks in complex scenarios.

Key words: cross-view image geo-localization, directional misalignment, ranking aggregation, directional robustness

中图分类号:

P237

盛怡宁, 赵理君, 张正, 唐娉. 基于排名聚合的跨视角图像地理定位方向鲁棒性增强方法^*[J]. 中国科学院大学学报, DOI: 10.7523/j.ucas.2025.019.

SHENG Yining, ZHAO Lijun, ZHANG Zheng, TANG Ping. A ranking aggregation-based method for enhancing directional robustness in cross-view image geo-localization[J]. Journal of University of Chinese Academy of Sciences, DOI: 10.7523/j.ucas.2025.019.

参考文献

[1] Lin T Y, Belongie S, Hays J.Cross-view image geolocalization[C]//2013 IEEE Conference on Computer Vision and Pattern Recognition. June 23-28, 2013, Portland, OR, USA. IEEE, 2013: 891-898. DOI: 10.1109/CVPR.2013.120.
[2] Downes L M, Kim D K, Steiner T J, et al.City-wide street-to-satellite image geolocalization of a mobile ground agent[C]//2022 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS). October 23-27, 2022, Kyoto, Japan. IEEE, 2022: 11102-11108. DOI: 10.1109/IROS47612.2022.9981996.
[3] McManus C, Churchill W, Maddern W, et al. Shady dealings: robust, long-term visual localisation using illumination invariance[C]//2014 IEEE International Conference on Robotics and Automation (ICRA). May 31 - June 7, 2014, Hong Kong, China. IEEE, 2014: 901-906. DOI: 10.1109/ICRA.2014.6906961.
[4] Häne C, Heng L, Lee G H, et al.3D visual perception for self-driving cars using a multi-camera system: calibration, mapping, localization, and obstacle detection[J]. Image and Vision Computing, 2017, 68: 14-27. DOI: 10.1016/j.imavis.2017.07.003.
[5] Middelberg S, Sattler T, Untzelmann O, et al.Scalable 6-DOF localization on mobile devices[C]// Computer Vision - ECCV 2014. Cham: Springer International Publishing, 2014: 268-283. DOI: 10.1007/978-3-319-10605-2_18.
[6] Workman S, Souvenir R, Jacobs N.Wide-area image geolocalization with aerial reference imagery[C]//2015 IEEE International Conference on Computer Vision (ICCV). December 7-13, 2015, Santiago, Chile. IEEE, 2015: 3961-3969. DOI: 10.1109/ICCV.2015.451.
[7] Liu L, Li H D.Lending orientation to neural networks for cross-view geo-localization[C]//2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). June 15-20, 2019, Long Beach, CA, USA. IEEE, 2019: 5617-5626. DOI: 10.1109/CVPR.2019.00577.
[8] Zhu S J, Shah M, Chen C.TransGeo: transformer is all you need for cross-view image geo-localization[C]//2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). June 18-24, 2022, New Orleans, LA, USA. IEEE, 2022: 1152-1161. DOI: 10.1109/CVPR52688.2022.00123.
[9] Zhu Y Y, Yang H J, Lu Y X, et al. Simple, effective and general: a new backbone for cross-view image geo-localization[EB/OL]. arXiv:2302.01572.(2023-02-03)[ 2025-01-17] https://arxiv.org/abs/2302.01572v1.
[10] Deuser F, Habel K, Oswald N.Sample4Geo: hard negative sampling for cross-view geo-localisation[C]//2023 IEEE/CVF International Conference on Computer Vision (ICCV). October 1-6, 2023, Paris, France. IEEE, 2023: 16801-16810. DOI: 10.1109/ICCV51070.2023.01545.
[11] Zhang X H, Li X Y, Sultani W, et al.GeoDTR+: toward generic cross-view geolocalization via geometric disentanglement[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2024, 46(12): 10419-10433. DOI: 10.1109/TPAMI.2024.3443652.
[12] Mi L, Xu C, Castillo-Navarro J, et al.ConGeo: robust cross-view geo-localization across ground view variations[C]// Computer Vision - ECCV 2024. Cham: Springer Nature Switzerland, 2025: 214-230. DOI: 10.1007/978-3-031-72630-9_13.
[13] Zhang X H, Li X Y, Sultani W, et al.Cross-view geo-localization via learning disentangled geometric layout correspondence[J]. Proceedings of the AAAI Conference on Artificial Intelligence, 2023, 37(3): 3480-3488. DOI: 10.1609/aaai.v37i3.25457.

[1]	王寅达, 陈嘉辉, 彭玲, 李兆博, 杨丽娜. 高分辨率关系图卷积网络遥感语义分割方法[J]. 中国科学院大学学报, 2025, 42(1): 107-115.
[2]	刘扬, 方俊永, 刘学, 王潇, 张晓红. 基于滤光片转轮式多光谱相机的辐射校正[J]. 中国科学院大学学报, 2024, 41(5): 636-643.
[3]	谢婉蓉, 刘少创, 吴运佳, 张烁. 长江正源当曲流域夏季高寒湿地信息遥感分类提取[J]. 中国科学院大学学报, 2024, 41(5): 625-635.
[4]	马忠辉, 崔珍珍, 闫东川, 刘冰洁, 万意, 王志强, 张成龙. HJ-1A/CCD1传感器的长时间序列交叉定标^*[J]. 中国科学院大学学报, 0, (): 240418-.
[5]	潘超, 李凉海, 曹海翊, 赵一鸣, 袁逸飞, 韩晓爽. 基于密度与局部统计的单光子点云去噪方法[J]. 中国科学院大学学报, 2024, 41(2): 268-274.
[6]	李昊, 马友青, 张烁, 刘少创. “祝融号”火星车驶离着陆平台决策支持几何参数计算方法[J]. 中国科学院大学学报, 2024, 41(1): 81-87.
[7]	曾建顺, 吕炎杰, 覃驭楚. 基于多尺度语义先验的街景图像修复算法^*[J]. 中国科学院大学学报, 0, (): 6-6.
[8]	陈若男, 彭玲, 刘玉菲, 卫志超, 吕蓓茹, 陈德跃. 引入空间距离信息的城郊山区道路提取与应用[J]. 中国科学院大学学报, 2022, 39(5): 658-667.
[9]	方韩康, 张波, 陈卫荣, 吴樊, 王超. 高分三号SAR影像L1A级产品精处理方法[J]. 中国科学院大学学报, 2022, 39(5): 648-657.
[10]	张晓平, 高珊珊, 陈明星, 赵艳艳. 夜间灯光数据在城市化及其资源环境效应研究中的热点主题追踪[J]. 中国科学院大学学报, 2022, 39(4): 490-501.
[11]	陈忠媛, 高永年. 长江水系湖泊型流域生态用地长时间序列时空变化分析[J]. 中国科学院大学学报, 2022, 39(2): 172-184.
[12]	刘杉, 张风丽, 韦诗莹, 刘娜, 邵芸. 基于极化分解组合的SAR图像视觉优化和建筑物损毁评估[J]. 中国科学院大学学报, 2020, 37(6): 750-759.
[13]	王帅, 覃驭楚, 杨富坤. 大面积建成区建筑密度遥感制图产品的精度验证[J]. 中国科学院大学学报, 2020, 37(1): 83-92.
[14]	顾鹏程, 王世新, 周艺, 刘文亮, 尚明. 基于时间序列DMSP/OLS夜间灯光数据的GDP预测模型[J]. 中国科学院大学学报, 2019, 36(2): 188-195.
[15]	杨渊, 陈凤翔, 虢韬, 时磊, 王成, 罗鑫. 基于LiDAR点云特征和模型拟合的高压线提取[J]. 中国科学院大学学报, 2018, 35(5): 612-616.

基于排名聚合的跨视角图像地理定位方向鲁棒性增强方法^*

A ranking aggregation-based method for enhancing directional robustness in cross-view image geo-localization

PDF (PC)

可视化

摘要/Abstract

引用本文

使用本文

参考文献

相关文章 15

编辑推荐

Metrics

本文评价

访问统计

联系我们