-
摘要: 跨视图定位技术正经历从通用框架向面向自动驾驶的专有化架构深刻转变. 为系统梳理这一领域, 首先回顾其从粗略检索到精准3-DoF姿态估计的发展脉络. 进而, 提出一个从“输入数据、模型架构、评价指标、损失函数”四个核心维度出发的分析框架, 以此对主流方法进行系统性解构与归因, 揭示不同技术路径的设计权衡. 此外, 详细评述自动驾驶跨视图定位数据集, 并指出当前评估在泛化能力考量上的不足; 为此, 构建新的包含结构化道路与越野场景的CSC-Shanmao数据集, 作为评估模型环境适应性的新基准, 并在其上验证了多种代表性方法, 结果表明面向自动驾驶设计的方法在精度与鲁棒性上优势显著, 但同时也暴露出精度、效率与泛化性之间的突出矛盾. 最后, 总结该技术在轻量化部署、多模态融合可靠性及评估体系等方面面临的严峻挑战, 并展望轻量化设计、可信评估基准及主动感知等未来关键研究方向.Abstract: Cross-view localization techniques is undergoing a profound shift from general-purpose frameworks to specialized architectures tailored for autonomous driving. To systematically review this field, we first trace its evolution from coarse retrieval to precise 3-DoF pose estimation. We then propose an analytical framework centered on four core dimensions(input data, model architecture, evaluation metrics, and loss functions)to deconstruct and attribute mainstream methods systematically, revealing the design trade-offs among different technical approaches. Furthermore, we provide a detailed review of autonomous driving cross-view localization datasets and identify the current evaluation shortcomings regarding generalization capability. To address this, a new dataset named CSC-Shanmao, containing both structured road and off-road scenarios, is constructed as a novel benchmark for assessing model environmental adaptability. Various representative methods are validated on this dataset, demonstrating that approaches designed specifically for autonomous driving exhibit significant advantages in accuracy and robustness, while also revealing prominent trade-off among accuracy, efficiency, and generalization. Finally, we summarize the critical challenges this technique faces in lightweight deployment, reliability of multi-modal fusion, and evaluation systems, and discuss key future research directions focusing on lightweight design, trustworthy evaluation benchmarks, and active perception.
-
Key words:
- cross-view localization /
- autonomous driving /
- deep learning /
- computer vision
-
图 2 面向自动驾驶的跨视图定位方法的发展历程
Fig. 2 Evolution of cross-view localization methods for autonomous driving
图 3 面向自动驾驶的跨视图定位方法一览
Fig. 3 Overview of cross-view localization methods for autonomous driving
图 6 面向自动驾驶的跨视图定位方法性能对比
Fig. 6 Performance comparison of cross-view localization methods for autonomous driving
图 7 跨视图定位方法未来研究方向
Fig. 7 Future research directions of cross-view localization methods
表 1 CVUSA数据集上的跨视图定位方法性能对比
Table 1 Performance comparison of cross-view localization methods on the CVUSA dataset
下载: 导出CSV
表 2 CVACTval数据集上的跨视图定位方法性能对比
Table 2 Performance comparison of cross-view localization methods on the CVACTval dataset
下载: 导出CSV
表 3 KITTI-CVL数据集上不同实验条件下的跨视图定位方法性能对比
Table 3 Performance comparison of cross-view localization methods under different experimental settings on the KITTI-CVL dataset
方法 定位$\downarrow$(m) 横向$\uparrow$(%) 纵向$\uparrow$(%) 方向$\uparrow$(%) 方向$\downarrow$($^\circ$) 平均 中位 0.25 m 0.5 m 1 m 3 m 中位$\downarrow$ 0.25 m 0.5 m 1 m 3 m 中位$\downarrow$ 1$^\circ$ 2$^\circ$ 3$^\circ$ 4$^\circ$ 5$^\circ$ 平均 中位 方向先验噪声为± 15°, 横/纵位置先验噪声为± 5m SIBCL[52] — — 25.59 46.26 72.63 — 0.54 21.91 41.22 64.47 — 0.64 56.05 79.70 — 90.89 — — 0.85 DSM[55] — — 8.51 13.97 23.44 — 3.48 7.67 11.02 20.03 — 4.97 3.13 8.29 — 17.44 — — 12.73 CVML[58] — — 6.11 12.24 23.78 — 2.40 5.97 11.68 23.73 — 2.46 — — — — — — — LM[60] — — 16.51 32.05 57.65 — 0.83 7.14 14.11 27.41 — 2.01 29.83 53.41 — 76.51 — — 3.23 PureACL[64] — — 84.58 99.54 99.98 — 0.12 98.55 100.00 100.00 — 0.09 31.18 54.13 — 76.00 — 3.57 1.78 方向先验噪声为± 10°, 横/纵位置先验噪声为± 20 m(同区域测试) DSM[55] — — — — 10.12 30.67 — — — 4.08 12.01 — 3.58 — 13.81 — 24.44 — — HC-Net[59] 0.80 0.50 — — 99.01 — — — — 92.20 — — 91.35 — — — 99.84 0.45 0.33 LM[60] 12.08 11.42 — — 35.54 70.77 — — — 5.22 15.88 — 19.64 — 51.76 — 71.72 3.72 2.83 BoostAcc[63] 10.01 5.19 — — 76.44 96.34 — — — 23.54 50.57 — 99.10 — 100.00 — 100.00 0.55 0.42 PureACL[64] 2.42 0.42 — — 91.95 93.40 — — — 91.86 92.12 — 32.17 — 53.89 — 71.66 3.97 1.77 SliceMatch[65] 7.96 4.39 — — 49.09 — — — — 15.19 — — 13.41 — — — 64.17 4.12 3.65 文献[66]中的方法 1.48 0.47 — — 95.47 — — — — 87.89 — — 89.40 — — — 99.31 0.49 0.30 CCVPE[67] 1.22 0.62 — — 97.35 98.65 — — — 77.13 96.08 — 77.39 — 99.47 — 99.95 0.67 0.54 T2GA[68] 0.20 0.17 — — 99.97 99.97 — — — 99.97 99.97 — 50.95 — 83.97 — 95.45 1.53 0.93 C2F-CCPE[78] 0.92 0.44 — — 98.52 99.02 — — — 88.97 96.77 — 99.97 — 99.99 — 100.00 0.01 0.01 FG2[88] 0.75 0.52 — — 99.73 — — — — 86.99 — — 61.17 — — — 95.65 1.28 0.74 CVLGSA[91] 1.15 0.35 — — 98.22 99.04 — — — 92.47 96.13 — 99.86 — 100.00 — 100.00 0.28 0.23 CPC-CVPE[93] 1.03 0.55 — — 97.87 — — — — 86.21 — — 86.21 — — — 97.98 1.01 0.65 方向先验噪声为± 10°, 横/纵位置先验噪声为± 20 m(跨区域测试) DSM[55] — — — — 10.77 31.37 — — — 3.87 11.73 — 3.53 — 14.09 — 23.95 — — HC-Net[59] 8.47 4.57 — — 75.00 — — — — 58.93 — — 33.58 — — — 83.78 3.22 1.63 LM[60] 12.58 12.11 — — 27.82 59.79 — — — 5.75 16.36 — 18.42 — 49.72 — 71.00 3.95 3.03 BoostAcc[63] 13.01 9.06 — — 57.72 86.77 — — — 14.15 34.59 — 98.98 — 100.00 — 100.00 0.56 0.43 PureACL[64] 6.20 0.61 — — 67.24 87.26 — — — 64.81 73.63 — 23.45 — 48.69 — 59.24 4.26 2.48 SliceMatch[65] 13.50 9.77 — — 32.43 — — — — 8.30 — — 46.82 — — — 46.82 4.20 6.61 文献[66]中的方法 7.97 3.52 — — 54.19 — — — — 23.10 — — 43.44 — — — 89.31 2.17 1.21 CCVPE[67] 9.16 3.33 — — 44.06 81.72 — — — 23.08 52.85 — 57.72 — 92.34 — 96.19 1.55 0.84 T2GA[68] 0.21 0.18 — — 99.92 99.96 — — — 99.91 99.91 — 38.31 — 80.12 — 92.48 2.04 1.38 C2F-CCPE[78] 8.16 3.14 — — 49.55 87.39 — — — 20.96 55.20 — 98.85 — 98.98 — 99.02 0.26 0.01 FG2[88] 7.45 4.03 — — 89.46 — — — — 12.42 — — 30.34 — — — 81.17 3.33 1.88 CVLGSA[91] 9.50 4.34 — — 60.00 91.24 — — — 21.36 44.68 — 99.53 — 100.00 — 100.00 0.31 0.24 方向先验噪声为± 45°, 横/纵位置先验噪声为± 20 m(同区域测试) SIBCL[52] — — 17.01 33.34 61.88 — 0.70 20.68 31.40 55.58 — 0.86 35.50 55.04 — 71.06 — 8.90 1.69 BoostAcc[63] — — 21.92 41.51 67.72 — 0.63 5.17 9.38 16.54 — 5.42 18.45 35.25 — 62.20 — 3.88 2.98 PureACL[64] — — 58.75 66.34 70.60 — 0.23 45.65 68.23 74.87 — 0.37 37.99 52.57 — 64.13 — 8.13 2.80 CCVPE[67] — — 11.05 22.05 42.09 — 1.22 7.45 13.36 24.97 — 3.35 34.66 60.72 — 84.03 — 3.43 1.53 T2GA[68] — — 76.20 97.73 99.97 — 0.14 98.18 99.89 99.97 — 0.07 48.97 66.29 — 90.54 — 2.19 1.07 方向先验噪声为± 45°, 横/纵位置先验噪声为± 20 m(跨区域测试) SIBCL[52] — — 16.90 32.62 58.24 — 0.71 19.58 30.74 49.25 — 1.05 25.02 50.18 — 60.16 — 9.67 1.96 BoostAcc[63] — — 14.88 29.20 52.64 — 0.93 3.38 5.56 11.73 — 8.92 12.46 24.87 — 47.26 — 5.61 4.26 PureACL[64] — — 50.25 62.63 64.56 — 0.25 7.41 62.04 64.23 — 0.44 19.95 36.25 — 56.49 — 10.96 3.16 CCVPE[67] — — 4.96 10.05 38.12 — 2.79 3.13 6.14 12.45 — 8.16 3.38 10.24 — 17.41 — 19.46 15.39 T2GA[68] — — 75.96 97.47 99.97 — 0.15 97.49 99.87 99.97 — 0.07 36.10 63.08 — 86.18 — 2.88 1.44
下载: 导出CSV
表 4 面向自动驾驶的跨视图定位方法分类(依据评价指标)
Table 4 Taxonomy of cross-view localization methods for autonomous driving based on evaluation metrics
模型评价指标 方法 时间及论文来源 输入 创新点 Top-K召回率 DSM[55] 2020 CVPR 地面+鸟瞰/卫星 极坐标变换对齐航拍图像与地面图像, 两流卷积网络学习深度特征 CVLNet[61] 2022 ACCV 地面+鸟瞰/卫星 通过几何驱动模块和场景先验匹配机制确定车载相机相对于卫星图中心的位置 Accurate3-DoF[62] 2022 TPAMI 地面+鸟瞰/卫星 设计双孪生网络结构, 提出两阶段定位机制, 使网络聚焦特定区域 HADGEO[79] 2024 ICASSP 地面+鸟瞰/卫星 设计新的损失函数, 结合全可学习卷积网络(ALFCN)提取特征 GeoSSK[89] 2025 MMM 地面+鸟瞰/卫星 设计基于交叉注意力的跨融合交互模块(CIM)和语义相似性知识蒸馏(SSKD) DSTG[90] 2025 TGRS 地面+鸟瞰/卫星 提出一种在不同分辨率级别之间进行教师模型与学生模型特征对齐的方法 定位与方向估计的误差及召回率 SIBCL[52] 2023 ICRA 地面+点云+卫星 用地面图像和点云估计车辆的3-DoF姿态 MapLocNet[53] 2024 IROS 地面+鸟瞰/卫星 使用从粗到精的特征注册策略进行视觉重定位 AutoVision[54] 2019 ICRA 地面+鸟瞰/卫星 用粒子滤波方法结合CVM-Net进行跨视图匹配 SLAM-G2S-Fusion[57] 2024 ICRA 地面+鸟瞰/卫星 将vSLAM与G2S方法结合形成一种从粗到细的选择方法 CVML[58] 2022 ECCV 地面+鸟瞰/卫星 利用密集的卫星描述符和相似性匹配来生成密集概率分布 HC-Net[59] 2023 NeurIPS 地面+鸟瞰/卫星 提出单应性估计模块消除重复采样并忽略不可观察内容 LM[60] 2022 CVPR 地面+鸟瞰/卫星 通过几何投影模块, 将卫星图像的深度特征投影到地面视角下, 再利用LM模块细化相机的姿态估计 BoostAcc[63] 2023 ICCV 地面+鸟瞰/卫星 提出几何引导交叉视图变换器, 解决了对象模糊和相机倾斜的问题 PureACL[64] 2023 ICCV 地面+鸟瞰/卫星 检测一致性关键点以及深度特征, 减少纯视觉匹配的模糊性 SliceMatch[65] 2023 CVPR 地面+鸟瞰/卫星 将水平视场分割成多个切片, 学习与定位和方向估计相关的特征 文献[66]中的方法 2023 NeurIPS 地面+鸟瞰/卫星 通过密集像素级流场来估计车载相机的3-DoF姿态 CCVPE[67] 2023 TU Delft 地面+鸟瞰/卫星 构建方向感知的图像描述符, 实现位置和方向的联合估计 T2GA[68] 2022 CVPR 地面+鸟瞰/卫星 T2GA模块整合离地特征, CycDA损失确保特征一致性, 等距投影损失(ERP)平衡关键点影响 BEVRender[70] 2024 IROS 地面+鸟瞰/卫星 生成局部鸟瞰图图像并将其与航拍视图对齐 Geo-tracking[71] 2022 IROS 地面+点云+卫星 将车载摄像头和激光雷达数据与地理配准的正射影像对齐 文献[74]中的方法 2024 JSTARS 点云+ 卫星 引入道路结构作为不同模态之间的桥梁 AGL-NET[75] 2024 IROS 点云+卫星 利用尺度分类器和特征差值方法处理尺度差异 C2F-CCPE[78] 2024 ICME 地面+鸟瞰/卫星 通过多尺度特征融合模块(LOFFM) 同时进行方向和位置的精确预测 FG2[88] 2025 CVPR 地面+鸟瞰/卫星 利用几何感知注意力与高度选择机制实现细粒度特征匹配 CVLGSA[91] 2025 IROS 地面+鸟瞰/卫星 通过视角驱动的注意力融合模块和投影稳定补丁引导优化器 CPC-CVPE[93] 2025 Sci. Data 地面+点云+卫星 通过多传感器融合实现精确的GNSS独立定位 其他 OrienterNet[76] 2023 CVPR 地面+ 卫星 用卷积神经网络提取图像语义特征并将其提升为鸟瞰图表示, 同时将开放街图编码为神经地图
下载: 导出CSV
表 5 传统跨视图定位数据集
Table 5 Traditional cross-view localization dataset
数据集 发布时间 图像分辨率 图像类型 数据量 地点 CVUSA[2] 2015 1232 × 224 (地面),
750 × 750 (卫星)地面图像和卫星图像 44416 对地面和卫星图像美国 San Francisco[57] 2015 256 × 256 地面图像和卫星图像 278561 幅地面图像和174217 幅卫星图像旧金山弗朗西斯科 CVACT[51] 2019 多尺寸 地面图像和卫星图像 44416 对地面和卫星图像以及92802 幅额外测试图像澳大利亚堪培拉 University- 1652 [10]2020 512 × 512 地面图像、无人机图像和卫星图像 1652 组地面、无人机和卫星图像世界各地72所大学 VIGOR[14] 2021 2048 × 1024 (地面),
640 × 640 (卫星)地面图像和卫星图像 238696 幅地面图像和90618 幅卫星图像纽约市、旧金山弗朗西斯科、
芝加哥和西雅图SUES-200[83] 2023 1080 ×1080 (无人机),
512 × 512 (卫星)无人机图像和卫星图像 24120 对无人机和卫星图像上海工程科技大学周围
下载: 导出CSV
表 6 面向自动驾驶的跨视图定位数据集
Table 6 Cross-view localization datasets for autonomous driving
数据集 发布时间 图像分辨率 图像类型 数据量 地点 KITTI-CVL[52] 2023 1242 × 375(地面),1280 ×1280 (卫星)地面图像和卫星图像 30973 幅地面视角图像和2750 幅卫星图像德国卡尔斯鲁厄市 FordAV-CVL[52] 2023 1656 × 860(地面),1280 ×1280 (卫星)地面图像和卫星图像 30209 对地面视角和
卫星图像美国密歇根州 Oxford RobotCar(+)[57] 2022 1280 × 960(地面),
600 × 600(卫星)地面图像和卫星图像 24000 对地面视角和
卫星图像英国牛津 KITTI(+)[60] 2023 1242 × 375(地面),1280 ×1280 (卫星)地面图像和卫星图像 30973 幅地面视角图像和2750 幅卫星图像德国卡尔斯鲁厄市 Ford Multi-AV[60] 2023 1656 × 860 (地面),1280 ×1280 (卫星)地面图像和卫星图像 30209 对地面视角和
卫星图像美国密歇根州 CSC-Shanmao
(本文数据集)2025 960 × 540(地面), 1280 ×1280 (卫星)地面图像和卫星图像 24115 对地面视角和
卫星图像中国湖南省长沙市以及
内蒙古包头市
下载: 导出CSV
表 7 现有先进方法在CSC-Shanmao数据集上的性能评测
Table 7 Benchmarking state-of-the-art methods on the CSC-Shanmao dataset
测试条件与方法 定位$\downarrow$(m) 横向$\uparrow$(%) 纵向$\uparrow$(%) 方向$\downarrow$($^\circ$) 方向$\uparrow$(%) 均值 中值 1 m 3 m 5 m 1 m 3 m 5 m 均值 中值 1$^\circ$ 3$^\circ$ 5$^\circ$ $\pm 10^\circ$, 结构化数据 LM[60] 8.69 4.96 24.74 57.93 65.95 31.21 62.07 76.55 1.68 1.25 42.16 83.45 95.09 文献[66]中的方法 16.06 15.61 0.66 4.45 16.12 4.59 14.59 25.43 3.98 3.31 15.40 46.26 69.74 CCVPE[67] 5.99 1.58 73.02 88.62 90.43 47.16 71.63 78.53 5.99 5.64 6.98 22.93 43.19 FG2[88] 8.09 7.58 16.49 50.46 76.12 8.62 26.67 42.27 7.80 7.00 5.66 18.28 33.45 $\pm 10^\circ$, 非结构化数据 LM[60] 4.77 3.86 31.47 56.70 81.23 41.48 66.08 77.10 2.46 1.56 41.64 66.40 83.44 文献[66]中的方法 20.23 19.20 2.44 8.20 16.51 3.48 11.28 18.50 4.84 4.20 11.75 36.14 58.04 CCVPE[67] 35.59 37.69 0.03 0.07 0.09 0.02 0.07 0.11 7.96 8.48 0.08 0.21 0.31 FG2[88] 8.82 7.31 13.88 41.18 63.38 9.35 31.52 50.36 10.32 9.19 4.20 12.35 23.50
下载: 导出CSV
-
[1] Lin T Y, Cui Y, Belongie S, et al. Learning deep representations for ground-to-aerial geolocalization. In: Proceedings of the IEEE conference on computer vision and pattern recognition. Boston, MA, USA: IEEE, 2015. 5007-5015 [2] Workman S, Jacobs N. On the location dependence of convolutional neural network features. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition Workshops. Boston, MA, USA: IEEE, 2015. 70-78 [3] Russakovsky O, Deng J, Su H, et al. Imagenet large scale visual recognition challenge. International journal of computer vision, 2015, 115: 211−252 doi: 10.1007/s11263-015-0816-y [4] Zhou B, Lapedriza A, Xiao J, et al. Learning deep features for scene recognition using places database. Advances in neural information processing systems, 2014, 27: 487 doi: 10.1109/icicct.2018.8473136 [5] Bromley J, Guyon I, LeCun Y, et al. Signature verification using a "siamese" time delay neural network. Advances in neural information processing systems, 1993, 07(4): 669−669 doi: 10.1142/9789812797926_0003 [6] Tian Y, Chen C, Shah M. Cross-view image matching for geo-localization in urban environments. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. Honolulu, HI, USA: IEEE, 2017. 1998-2006 [7] Vo N N, Hays J. Localizing and orienting street views using overhead imagery. In: Proceedings of the European conference on computer vision. Cham: Springer International Publishing, 2016. 494-509 [8] Hu S, Feng M, Nguyen R M H, et al. Cvm-net: Cross-view matching network for image-based ground-to-aerial geo-localization. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. Salt Lake City, UT, USA: IEEE, 2018. 7258-7267 [9] Arandjelovic R, Gronat P, Torii A, et al. NetVLAD: CNN architecture for weakly supervised place recognition. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2016, 40(6): 1437−1451 doi: 10.1109/cvpr.2016.572 [10] Zheng Z, Wei Y, Yang Y. University-1652: A multi-view multi-source benchmark for drone-based geo-localization. In: Proceedings of the 28th ACM international conference on Multimedia. Napoli: 2020. 1395-1403 [11] Zheng Z, Zheng L, Garrett M, et al. Dual-path convolutional image-text embeddings with instance loss. ACM Transactions on Multimedia Computing, Communications, and Applications (TOMM), 2020, 16(2): 1−23 [12] Sun B, Chen C, Zhu Y, et al. Geocapsnet: Ground to aerial view image geo-localization using capsule network. In: Proceedings of the 2019 IEEE International Conference on Multimedia and Expo (ICME). Shanghai, China: IEEE, 2019. 742-747 [13] Zhu Y, Sun B, Lu X, et al. Geographic semantic network for cross-view image geo-localization. IEEE Transactions on Geoscience and Remote Sensing, 2022, 60: 1−15 [14] Zhu S, Yang T, Chen C. Vigor: Cross-view image geo-localization beyond one-to-one retrieval. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. IEEE, 2020. 5316-5325 [15] Guo Y, Choi M, Li K, et al. Soft exemplar highlighting for cross-view image-based geo-localization. IEEE transactions on image processing, 2022, 31: 2094−2105 doi: 10.1109/TIP.2022.3152046 [16] Gu B, Ling H, Shi Y, et al. Mutual relative position learning transformer for cross-view geo-localization. In: Proceedings of the 2023 IEEE International Conference on Image Processing (ICIP). Kuala Lumpur, Malaysia: IEEE, 2023. 286-290 [17] Xia Z. Ground-to-aerial image matching for vehicle localization[Ph.D. dissertation], Delft University of Technology, Netherlands, 2024 [18] Park J, Sung C, Lee S, et al. Cross-view geo-localization via effective negative sampling. In: Proceedings of the 2024 24th International Conference on Control, Automation and Systems (ICCAS). Jeju, Korea: IEEE, 2024. 1078-1083 [19] Zhai M, Bessinger Z, Workman S, et al. Predicting ground-level scene layout from aerial imagery. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. Honolulu, HI, USA: IEEE, 2017. 4132-4140 [20] Cai S, Guo Y, Khan S, et al. Ground-to-aerial image geo-localization with a hard exemplar reweighting triplet loss. In: Proceedings of the IEEE/CVF international conference on computer vision. Seoul, Korea (South): IEEE, 2019. 8390-8399 [21] Shi Y, Liu L, Yu X, et al. Spatial-aware feature aggregation for cross-view image based geo-localization. In: Proceedings of the 33rd Annual Conference on Neural Information Processing Systems, NeurIPS 2019. Vancouver, British Columbia, Canada: Curran Associates, 2019. 10090-10100 [22] Wang J, Li X. Cross-attention network for cross-view image geo-localization. In: Proceedings of the 2023 6th International Symposium on Autonomous Systems (ISAS). Nanjing, China: IEEE, 2023. 1-6 [23] Xia P, Wan Y, Zheng Z, et al. Enhancing cross-view geo-localization with domain alignment and scene consistency. IEEE Transactions on Circuits and Systems for Video Technology, 2024, 34(12): 13271−13281 doi: 10.1109/TCSVT.2024.3443510 [24] Zhang X, Li X, Sultani W, et al. Cross-view geo-localization via learning disentangled geometric layout correspondence. In: Proceedings of the AAAI conference on artificial intelligence. Washington, DC, USA: AAAI Press, 2023. 37: 3480-3488 [25] Zhang X, Li X, Sultani W, et al. GeoDTR+: toward generic cross-view geolocalization via geometric disentanglement. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2024, 46(12): 10419−10433 doi: 10.1109/TPAMI.2024.3443652 [26] Wang T, Zheng Z, Yan C, et al. Each part matters: Local patterns facilitate cross-view geo-localization. IEEE Transactions on Circuits and Systems for Video Technology, 2021, 32(2): 867−879 doi: 10.1109/tcsvt.2021.3061265 [27] Yagi R, Iwasaki A. Lightweight CNN for cross-view geo-localization using aerial image. In: Proceedings of the IGARSS 2023 - 2023 IEEE International Geoscience and Remote Sensing Symposium. Pasadena, CA, USA: IEEE, 2023. 6266-6269 [28] Ahn W J, Park S Y, Pae D S, et al. Bridging viewpoints in cross-view geo-localization with Siamese vision transformer. IEEE Transactions on Geoscience and Remote Sensing, 2024, 62: 1−12 doi: 10.1109/tgrs.2024.3429570 [29] Ge F, Zhang Y, Wang L, et al. Joint representation learning based on feature center region diffusion and edge radiation for cross-view geo-localization. IEEE Transactions on Geoscience and Remote Sensing, 2025, 63: 1−13 doi: 10.1109/tgrs.2024.3515484 [30] Lin J, Zheng Z, Zhong Z, et al. Joint representation learning and keypoint detection for cross-view geo-localization. IEEE Transactions on Image Processing, 2022, 31: 3780−3792 doi: 10.1109/TIP.2022.3175601 [31] Rodrigues R, Tani M. Semgeo: Semantic keywords for cross-view image geo-localization. In: Proceedings of the ICASSP 2023-2023 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP). Rhodes Island, Greece: IEEE, 2023. 1-5 [32] Wang T, Zheng Z, Zhu Z, et al. Learning cross-view geo-localization embeddings via dynamic weighted decorrelation regularization. IEEE Transactions on Geoscience and Remote Sensing, 2024, 62: 1−12 doi: 10.1109/tgrs.2024.3491757 [33] Hu D, Yuan X, Xi H, et al. Road structure inspired UGV-datellite cross-view geo-localization. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2024, 17: 16767−16786 doi: 10.1109/JSTARS.2024.3457756 [34] Rodrigues R, Tani M. Are these from the same place? seeing the unseen in cross-view image geo-localization. In: Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision. Waikoloa, HI, USA: IEEE, 2021. 3752-3760 [35] Yan Y, Wang M, Su N, et al. IML-Net: A framework for cross-view geo-localization with multi-domain remote sensing data. Remote Sensing, 2024, 16(7): 1249 doi: 10.3390/rs16071249 [36] Regmi K, Shah M. Bridging the domain gap for ground-to-aerial image matching. In: Proceedings of the IEEE/CVF International Conference on Computer Vision. Seoul, Korea: IEEE, 2019. 470-479 [37] Isola P, Zhu J Y, Zhou T, et al. Image-to-image translation with conditional adversarial networks. In: Proceedings of the IEEE conference on computer vision and pattern recognition. Honolulu, HI, USA: IEEE, 2017. 1125-1134 [38] Zhang H, Xu T, Li H, et al. Stackgan: Text to photo-realistic image synthesis with stacked generative adversarial networks. In: Proceedings of the IEEE international conference on computer vision. Venice, Italy: IEEE, 2017. 5907-5915 [39] Toker A, Zhou Q, Maximov M, et al. Coming down to earth: Satellite-to-street view synthesis for geo-localization. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. Nashville, TN, USA: IEEE, 2021. 6488-6497 [40] Tian X, Shao J, Ouyang D, et al. Smdt: Cross-view geo-localization with image alignment and transformer. In: Proceedings of the IEEE International Conference on Multimedia and Expo (ICME). Taipei, Taiwan: IEEE, 2022. 1-6 [41] Yang H, Lu X, Zhu Y. Cross-view geo-localization with layer-to-layer transformer. Advances in Neural Information Processing Systems, 2021, 34: 29009−29020 [42] Dai M, Hu J, Zhuang J, et al. A transformer-based feature segmentation and region alignment method for UAV-view geo-localization. IEEE Transactions on Circuits and Systems for Video Technology, 2021, 32(7): 4376−4389 doi: 10.1109/tcsvt.2021.3135013 [43] Zhu S, Shah M, Chen C. Transgeo: Transformer is all you need for cross-view image geo-localization. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. New Orleans, LA, USA: IEEE, 2022. 1152-1161 [44] Wang Y, Xia Y, Lu T, et al. An efficient method based on multi-view semantic alignment for cross-view geo-localization. In: Proceedings of the 2023 International Joint Conference on Neural Networks (IJCNN). Gold Coast, Australia: IEEE, 2023. 1-8 [45] Zhang W, Li J, Chen H, et al. MT-GEO: A multi-scale feature extraction network for cross-view geo-localization between street-view and remote sensing imagery. In: Proceedings of the IGARSS 2024-2024 IEEE International Geoscience and Remote Sensing Symposium. Athens, Greece: IEEE, 2024. 6964-6968 [46] Zhang W, Zhong Z, Chen H, et al. GeoViewMatch: A multi-scale feature-matching network for cross-view geo-localization using swin-transformer and contrastive learning. Remote Sensing, 2024, 16(4): 678 doi: 10.3390/rs16040678 [47] Li G, Qian M, Xia G S. Unleashing unlabeled data: A paradigm for cross-view geo-localization. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. Seattle, WA, USA: IEEE, 2024. 16719-16729 [48] Shore T, Mendez O, Hadfield S. PEnG: Pose-enhanced geo-localisation. IEEE Robotics and Automation Letters, 2025, 10(4): 3835−3842 doi: 10.1109/LRA.2025.3546513 [49] 周博文, 李阳, 马鑫骥, 苗壮, 张睿. 深度学习的跨视角地理定位方法综述. 中国图象图形学报, 2024, 29(12): 3543−3563 doi: 10.11834/jig.230858Zhou Bo-Wen, Li Yang, Ma Xin-Ji, Miao Zhuang, Zhang Rui. A survey of cross-view geo-localization methods based on deep learning. Journal of Image and Graphics, 2024, 29(12): 3543−3563 doi: 10.11834/jig.230858 [50] 刘楝, 李江杉, 许经乾, 等. 跨视角图像地理定位技术研究综述. 智能安全, 2023, 2(01): 53−61 doi: 10.12407/j.issn.2097-2075.2023.01.053Liu Lian, Li Jiang-Shan, Xu Jing-Qian, et al. Cross-view image geo-localization review. Artificial Intelligence Security, 2023, 2(01): 53−61 doi: 10.12407/j.issn.2097-2075.2023.01.053 [51] Liu L, Li H. Lending orientation to neural networks for cross-view geo-localization. In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. Long Beach, CA, USA: IEEE, 2019. 5624-5633 [52] Wang S, Zhang Y, Vora A, et al. Satellite image based cross-view localization for autonomous vehicle. In: Proceedings of the 2023 IEEE International Conference on Robotics and Automation (ICRA). London, United Kingdom: IEEE, 2023. 3592-3599 [53] Wu H, Zhang Z, Lin S, et al. Maplocnet: Coarse-to-fine feature registration for visual re-localization in navigation maps. In: Proceedings of the IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS). Abu Dhabi, United Arab Emirates: IEEE, 2024. 13198-13205 [54] Heng L, Choi B, Cui Z, et al. Project autovision: Localization and 3d scene perception for an autonomous vehicle with a multi-camera system. In: Proceedings of the 2019 International Conference on Robotics and Automation (ICRA). Montreal, QC, Canada: IEEE, 2019. 4695-4702 [55] Shi Y, Yu X, Campbell D, et al. Where am i looking at? joint location and orientation estimation by cross-view matching. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. Seattle, WA, USA: IEEE, 2020. 4063-4071 [56] Dixit D, Verma S, Tokekar P. Evaluation of cross-view matching to improve ground vehicle localization with aerial perception. arXiv preprint arXiv: 2003.06515, 2020. [57] Zhang Y, Shi Y, Wang S, et al. Increasing SLAM pose accuracy by ground-to-satellite image registration. In: Proceedings of the IEEE International Conference on Robotics and Automation (ICRA). Yokohama, Japan: IEEE, 2024. 8522-8528 [58] Xia Z, Booij O, Manfredi M, et al. Visual cross-view metric localization with dense uncertainty estimates. In: Proceedings of the European Conference on Computer Vision. Tel Aviv, IsraelP: Springer Nature Switzerland, 2022. 90-106 [59] Wang X, Xu R, Cui Z, et al. Fine-grained cross-view geo-localization using a correlation-aware homography estimator. Advances in Neural Information Processing Systems, 2023, 36: 5301−5319 [60] Shi Y, Li H. Beyond cross-view image retrieval: Highly accurate vehicle localization using satellite image. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. New Orleans, LA, USA: IEEE, 2022. 16989-16999 [61] Shi Y, Yu X, Wang S, et al. Cvlnet: cross-view semantic correspondence learning for video-based camera localization. In: Proceedings of the Asian Conference on Computer Vision. Osaka, Japan: Springer Nature Switzerland, 2022. 123-141 [62] Shi Y, Yu X, Liu L, et al. Accurate 3-DoF camera geo-localization via ground-to-satellite image matching. IEEE transactions on pattern analysis and machine intelligence, 2022, 45(3): 2682−2697 doi: 10.1109/tpami.2022.3189702 [63] Shi Y, Wu F, Perincherry A, et al. Boosting 3-dof ground-to-satellite camera localization accuracy via geometry-guided cross-view transformer. In: Proceedings of the IEEE/CVF International Conference on Computer Vision. Paris, France: IEEE, 2023. 21516-21526 [64] Wang S, Zhang Y, Perincherry A, et al. View consistent purification for accurate cross-view localization. In: Proceedings of the IEEE/CVF International Conference on Computer Vision. Paris, France: IEEE, 2023. 8163-8172 [65] Lentsch T, Xia Z, Caesar H, et al. Slicematch: Geometry-guided aggregation for cross-view pose estimation. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. Vancouver, BC, Canada: IEEE, 2023. 17225-17234 [66] Song Z, Lu J, Shi Y. Learning dense flow field for highly-accurate cross-view camera localization. Advances in Neural Information Processing Systems, 2023, 36: 70612−70625 doi: 10.52202/075280-3094 [67] Xia Z, Booij O, Kooij J F P. Convolutional cross-view pose estimation. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2023, 46(5): 3813−3831 [68] Wang S, Nguyen C, Liu J, et al. View from above: orthogonal-view aware cross-view localization. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. Seattle, WA, USA: IEEE, 2024. 14843-14852 [69] Zhang, Zhihuang, et al. Bev-locator: An end-to-end visual semantic localization network using multi-view images. Science China Information Sciences, 2025, 68(2): 1−17 doi: 10.1007/s11432-023-4114-6 [70] Jin, Lihong, et al. BEVRender: vision-based cross-view vehicle registration in off-road GNSS-denied environment. In: Proceedings of the 2024 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS). Abu Dhabi, United Arab Emirates: IEEE, 2024. 11032-11039 [71] Fervers F, Bullinger S, Bodensteiner C, et al. Continuous self-localization on aerial images using visual and lidar sensors. In: Proceedings of the 2022 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS). Kyoto, Japan: IEEE, 2022. 7028-7035 [72] Fervers F, Bullinger S, Bodensteiner C, et al. Uncertainty-aware vision-based metric cross-view geolocalization. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. Vancouver, BC, Canada: IEEE, 2023. 21621-21631 [73] Li L, Ma Y, Tang K, et al. Geo-localization with transformer-based 2D-3D match network. IEEE Robotics and Automation Letters, 2023, 8(8): 4855−4862 doi: 10.1109/LRA.2023.3290526 [74] Hu D, Yuan X, Xi H, et al. Road structure inspired UGV-satellite cross-view geo-localization. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2024, 17: 16767−16786 doi: 10.1109/JSTARS.2024.3457756 [75] Guan T, Xian R, Wang X, et al. AGL-NET: Aerial-ground cross-modal global localization with varying scales. In: Proceedings of the 2024 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS). Abu Dhabi, United Arab Emirates: IEEE, 2024. 8161-8161 [76] Sarlin P E, DeTone D, Yang T Y, et al. Orienternet: Visual localization in 2d public maps with neural matching. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. Vancouver, BC, Canada: IEEE, 2023. 21632-21642 [77] Guo X, Peng H, Hu J, et al. From satellite to ground: Satellite assisted visual localization with cross-view semantic matching. In: Proceedings of the 2024 IEEE International Conference on Robotics and Automation (ICRA). Yokohama, Japan: IEEE, 2024. 3977-3983 [78] Tang Y, Huang Q, Zhu Y. C2F-CCPE: Coarse-to-fine cross-view camera pose estimation. In: Proceedings of the 2024 IEEE International Conference on Multimedia and Expo (ICME). Niagara Falls, ON, Canada: IEEE, 2024. 1-6 [79] Li C, Yan C, Xiang X, et al. HADGEO: Image based 3-DoF cross-view geo-localization with hard sample mining. In: Proceedings of the ICASSP 2024-2024 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP). Seoul, Korea, Republic of: IEEE, 2024. 3520-3524 [80] Zhu J, Yan S, Cheng X, et al. CLaSP: Cross-view 6-DoF localisation assisted by synthetic panorama. IET Computer Vision, 2024, 18(7): 859−874 doi: 10.1049/cvi2.12285 [81] Miao J, Wen T, Luo Z, et al. Efficient end-to-end visual localization for autonomous driving with decoupled bev neural matching. In: Proceedings of the 2025 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS). Hangzhou, China: IEEE, 2025. 10719-10726. [82] Workman S, Jacobs N. On the location dependence of convolutional neural network features. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition Workshops. Boston, MA, USA: IEEE, 2015. 70-78 [83] Zhu R, Yin L, Yang M, et al. SUES-200: A multi-height multi-scene cross-view image benchmark across drone and satellite. IEEE Transactions on Circuits and Systems for Video Technology, 2023, 33(9): 4825−4839 doi: 10.1109/TCSVT.2023.3249204 [84] Geiger A, Lenz P, Stiller C, et al. Vision meets robotics: The kitti dataset. The international journal of robotics research, 2013, 32(11): 1231−1237 doi: 10.1177/0278364913491297 [85] Agarwal S, Vora A, Pandey G, et al. Ford multi-AV seasonal dataset. The International Journal of Robotics Research, 2020, 39(12): 1367−1376 doi: 10.1177/0278364920961451 [86] Maddern W, Pascoe G, Linegar C, et al. 1 year, 1000 km: The oxford robotcar dataset. The International Journal of Robotics Research, 2017, 36(1): 3−15 doi: 10.1177/0278364916679498 [87] Maddern W, Pascoe G, Gadd M, et al. Real-time kinematic ground truth for the oxford robotcar dataset. arXiv preprint arXiv: 2002.10152, 2020. [88] Xia Z, Alahi A. FG2: Fine-grained cross-view localization by fine-grained feature matching. In: Proceedings of the Computer Vision and Pattern Recognition Conference. Nashville, TN, USA: IEEE, 2025. 6362-6372 [89] Chen, Guanli, et al. Cross-view geo-localization via learning correspondence semantic similarity knowledge. In: Proceedings of the International Conference on Multimedia Modeling. Singapore: Springer Nature Singapore, 2025. 220-233 [90] Liang, Junxiong, et al. DSTG: Distillation swin transformer for cross-view geo-localization. IEEE Transactions on Geoscience and Remote Sensing, 2025, 63: 1−17 doi: 10.1109/tgrs.2025.3588220 [91] Qin, Riyu, et al. Focusing on projection-stable patch: cross-view localization with geometric-semantic alignment. In: Proceedings of the 2025 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS). Hangzhou, China: IEEE, 2025. 15438-15444 [92] Di ng, Xiaokun, et al. Cross-view geo-localization via 3D gaussian splatting-based novel view synthesis. Remote Sensing, 2025, 17(22): 3673 doi: 10.3390/rs17223673 [93] Yuan X, Wang K, Qin R, et al. A multi-modality ground-to-air cross-view pose estimation dataset for field robots. Scientific Data, 2025, 12(1): 754 doi: 10.1038/s41597-025-05075-9 [94] Durgam A, Paheding S, Dhiman V, et al. Cross-view geo-localization: a survey. IEEE Access, 2024, 12: 192028−192050 doi: 10.1109/ACCESS.2024.3507280 [95] Hadsell R, Chopra S, LeCun Y. Dimensionality reduction by learning an invariant mapping. In: Proceedings of the 2006 IEEE computer society conference on computer vision and pattern recognition (CVPR'06). New York, NY, USA: IEEE, 2006, 2: 1735-1742 [96] Girshick R. Fast r-cnn. In: Proceedings of the IEEE international conference on computer vision. Santiago, Chile: IEEE, 2015. 1440-1448 -
下载:
计量
- 文章访问数: 12
- HTML全文浏览量: 4
- 被引次数: 0
