Structure-Aware Matching for Robust UAV Thermal-to-Satellite Geo-Localization

Chun Liu; Yuxin Wang; Zhenxiang Huo; Linlin Zhu; Tao Xu; Henggang Han

cs.CVOct 9, 2026

Structure-Aware Matching for Robust UAV Thermal-to-Satellite Geo-Localization

Authors: Chun Liu, Yuxin Wang, Zhenxiang Huo, Linlin Zhu, Tao Xu, Henggang Han

Organizations: State Key Laboratory of Spatial Datum, Henan University, Zhengzhou 450046, China · College of Remote Sensing and Geoinformatics Engineering, Henan University, Zhengzhou 450046, China · School of Computer and Information Engineering, Henan University, Zhengzhou 450046, China · Henan Yunhuan NetLink UAV Technology Co., Ltd., Anyang 455000, China

Abstract

UAV thermal-to-satellite geo-localization localizes UAVs by matching real-time thermal images against georeferenced satellite images, enabling autonomous navigation in GNSS-denied and low-visibility conditions. However, substantial cross-modal disparities in color, brightness and viewpoint hinder conventional appearance-based methods from establishing reliable correspondences. While existing approaches predominantly address global modality shifts, they overlook the invariant scene structures inherently shared between thermal and satellite views. To bridge this gap, we propose SAMF, a Structure-Aware Matching Framework that explicitly exploits stable cross-modal structural cues to improve UAV thermal geolocalization. Specifically, the Directional Structure Enhancement Module (DSEM) strengthens robust structural representations by capturing directional and low-frequency spatial patterns, and the Modality-Aware Affine Remapping Module (MARM) subsequently applies affine correction to these enhanced features, further promoting cross-modal alignment. Additionally, we introduce the Thermal-UAV-Paired benchmark derived from the Thermal-UAV dataset. Extensive experiments on both the Boson-nighttime and Thermal-UAV-Paired datasets show that SAMF achieves state-of-the-art performance. Ablation studies further validate the effectiveness and synergistic complementarity of DSEM and MARM. The code and dataset are released at: https://github.com/YuxinWang879/SAMF.git.

Figures & tables

Fig. 1: Representative examples from the Boson-nighttime dataset. The left column displays UAV thermal images, while the right column presents the corresponding satellite reference images. Although thermal and satellite images captured at the same locations exhibit substantial visual appearance discrepancies, geometric structures such as surface stripe orientations, regional boundary contours, and coarse-grained spatial layouts exhibit strong cross-modal consistency. These consistent structural features provide potentially stable cues for cross-modal matching.

Fig. 2: The pipline of the proposed structure-aware matching framework for UAV thermal-to-satellite geo-localization. The thermal query image and satellite image are fed into a shared-parameter Swin Transformer backbone to extract feature maps. The backbone features are then enhanced by the Directional Structure Enhancement Module, where adaptive fusion of the semantic-preserving branch, multi-directional response branch, and low-frequency structure branch strengthens directional structures and low-frequency spatial organizations that are more stable under cross-modal conditions. Subsequently, the Modality-Aware Affine Remapping module performs lightweight channel-wise statistical correction on the structure-enhanced features, reducing the statistical distribution shift between thermal and satellite features.

Fig. 3: Typical scene examples from the Thermal-UAV-Paired dataset. The left column shows UAV thermal images, and the right column shows satellite reference images cropped according to the corresponding center locations. Unlike the desert scenes in Boson-nighttime dataset, this dataset contains more complex ground object types, such as sports fields, roads, and farmlands.

Dataset	Split	Paired Samples
Boson-nighttime	Training	10256
	Validation	13011
	Testing	26568
Thermal-UAV-Paired	Training	7886
	Validation	1407
	Testing	2350

TABLE I: Statistics of the Boson-nighttime and Thermal-UAV-Paired datasets.

Backbone baselines
Method	Venue	$\mathrm{R}@1$	$\mathrm{R}@5$	$R_{512}@1$	$R_{512}@5$	$L_{2}^{512}$ (m)
ResNet-18	–	63.2	75.6	80.0	89.5	24.6
ResNet-50	–	63.5	75.9	78.1	88.5	26.4
DINOv3-ConvNeXt-T	–	65.7	76.6	82.8	90.8	35.9
DINOv3-ConvNeXt-B	–	72.9	82.4	86.7	94.1	36.6
DINOv3-ViT-S	–	83.2	90.0	93.9	96.9	19.4

TABLE II: Comparison with existing methods on the Boson-nighttime test set. The best results are highlighted in bold .

Backbone baselines
Method	Venue	$\mathrm{R}@1$	$\mathrm{R}@5$	$R_{512}@1$	$R_{512}@5$	$L_{2}^{512}$ (m)
ResNet-18	–	93.1	98.0	94.9	99.0	12.8
ResNet-50	–	93.3	97.9	94.6	98.6	13.1
DINOv3-ConvNeXt-T	–	94.1	98.3	96.7	99.1	8.5
DINOv3-ConvNeXt-B	–	77.7	88.0	84.7	94.3	41.5
DINOv3-ViT-S	–	94.6	98.1	97.5	99.5	6.3

TABLE III: Comparison with existing methods on the Thermal-UAV-Paired test set. The best results are highlighted in bold .

Backbone	DSEM	MARM	$\mathrm{R}\text{@}1$	$\mathrm{R}\text{@}5$	$\mathrm{R}_{512}\text{@}1$	$\mathrm{R}_{512}\text{@}5$	$L_{512}^{2}~(\mathrm{m})$
Swin-T			84.1	90.9	93.7	97.7	26.5
Swin-T	$\checkmark$		82.8	91.7	93.4	98.2	25.5
Swin-T		$\checkmark$	83.3	91.1	93.3	97.7	29.4
Swin-T	$\checkmark$	$\checkmark$	93.6	97.4	98.2	99.6	12.3

TABLE IV: Ablation results of DSEM and MARM modules. The best results are highlighted in bold .

Backbone	DSEM			$\mathrm{R}\text{@}1$	$\mathrm{R}_{512}\text{@}1$	$L_{512}^{2}\,(\mathrm{m})$
	sem	dir	low
Swin-T	$\checkmark$	$\checkmark$	$\checkmark$	93.6	98.2	12.3
Swin-T	$\checkmark$		$\checkmark$	87.8	95.2	14.3
Swin-T		$\checkmark$	$\checkmark$	81.5	91.8	16.9
Swin-T	$\checkmark$	$\checkmark$		70.5	85.9	22.4

TABLE V: Ablation results of the three internal branches of DSEM. In the table, sem, dir, and low denote the semantic-preserving branch, directional response branch, and low-frequency structure branch of DSEM, respectively. The best results are highlighted in bold .

Backbone	Kernel	$\mathrm{R}\text{@}1$	$\mathrm{R}_{512}\text{@}1$	$L_{512}^{2}\,(\mathrm{m})$
Swin-T	3	82.2	92.5	19.2
Swin-T	5	93.6	98.2	12.3
Swin-T	7	85.5	94.2	13.0

TABLE VI: Parameter sensitivity analysis of the convolution kernel size in the low-frequency structure branch of DSEM. The best results are highlighted in bold .

Backbone	Num. Dir.	$\mathrm{R}\text{@}1$	$\mathrm{R}_{512}\text{@}1$	$L_{512}^{2}\,(\mathrm{m})$
Swin-T	4	87.4	94.8	13.4
Swin-T	8	93.6	98.2	12.3
Swin-T	12	90.5	96.7	11.5

TABLE VII: Parameter sensitivity analysis of the number of direction-sensitive units in DSEM. The best results are highlighted in bold .

Fig. 4: Stage-wise Grad-CAM visualization analysis of the DSEM and MARM modules. The first column shows the original input images, with the thermal query image in the upper row and the corresponding satellite image in the lower row. The remaining three columns show the Grad-CAM heatmaps generated from the Swin-T backbone feature $F$ , the DSEM-enhanced feature $F_{\mathrm{DSEM}}$ , and the MARM remapped feature $F_{\mathrm{MARM}}$ , respectively. The “cos” value between the two images in each column denotes the cosine similarity between the global descriptors of the thermal query image and the satellite image at the corresponding stage. A higher value indicates stronger matching consistency between the two images in the current feature space. Warm-colored regions represent spatial locations that contribute more to the current matching similarity.

Fig. 5: Visualization comparison of retrieval results on the test set. Each row represents a test sample. The first column displays the UAV thermal query image; the second and third columns show the Top-1 satellite images retrieved by the Swin baseline (red box) and the proposed method (green box), respectively. A red box indicates a mismatch with the ground-truth location, while a green box denotes a correct match. As observed, in low-semantic scenes such as desert stripes and indistinct roads, the baseline method is prone to mismatches due to appearance inconsistencies, often retrieving images with similar landforms but incorrect locations. In contrast, the proposed method achieves more stable cross-modal matching by leveraging directional structures and low-frequency spatial organization.

Cross-View Geo-Localization

Cross-view geo-localization matches ground, drone, or aerial images to maps or satellite views to estimate where they were captured.

See all 94 papers

SCC-Loc: A Unified Semantic Cascade Consensus Framework for UAV Thermal Geo-Localization
Apr 3, 2026Xiaoran Zhang, Yu Liu, Jinyu Liang +3Cross-View Geo-Localization GNSS-Denied Localization
College of Systems Engineering, National University of Defense Technology, Changsha 410073, China
ScalLoc: Continuous Oblique Thermal UAV Geo-Localization with a 2D Satellite Map
Oct 10, 2026Xiaoran Zhang, Yu Liu, Jinyu Liang +1Cross-View Geo-Localization GNSS-Denied Localization
College of Systems Engineering, National University of Defense Technology, Changsha 410073, China
Unifying UAV Cross-View Geo-Localization via 3D Geometric Perception
Apr 2, 2026Haoyuan Li, Wen Yang, Fang Xu +4Cross-View Geo-Localization GNSS-Denied Localization
School of Electronic Information, Wuhan University, Wuhan 430072, China · School of Artificial Intelligence, Wuhan University, Wuhan 430072, China · Technology and Engineering Center for Space Utilization and the Key Laboratory of Space Utilization, Chinese Academy of Sciences, Beijing 100094, China +1

Apr 3, 2026J/K move · Enter open · S save

SCC-Loc: A Unified Semantic Cascade Consensus Framework for UAV Thermal Geo-Localization

Xiaoran Zhang, Yu Liu, Jinyu Liang, Kangqiushi Li+2

College of Systems Engineering, National University of Defense Technology, Changsha 410073, China

Cross-View Geo-Localization GNSS-Denied Localization Cross-Modal Image Registration UAV Remote Sensing

Open paper AskPDF arXiv

Cross-modal Thermal Geo-localization (TG) provides a robust, all-weather solution for Unmanned Aerial Vehicles (UAVs) in Global Navigation Satellite System (GNSS)-denied environments. However, profound thermal-visible modality gaps introduce severe feature ambiguity, systematically corrupting conventional coarse-to-fine registration. To dismantle this bottleneck, we propose SCC-Loc, a unified Semantic-Cascade-Consensus localization framework. By sharing a single DINOv2 backbone across global retrieval and MINIMA

_{\text{RoMa}}

matching, it minimizes memory footprint and achieves zero-shot, highly accurate absolute position estimation. Specifically, we tackle modality ambiguity by introducing three cohesive components. First, we design the Semantic-Guided Viewport Alignment (SGVA) module to adaptively optimize satellite crop regions, effectively correcting initial spatial deviations. Second, we develop the Cascaded Spatial-Adaptive Texture-Structure Filtering (C-SATSF) mechanism to explicitly enforce geometric consistency, thereby eradicating dense cross-modal outliers. Finally, we propose the Consensus-Driven Reliability-Aware Position Selection (CD-RAPS) strategy to derive the optimal solution through a synergy of physically constrained pose optimization. To address data scarcity, we construct Thermal-UAV, a comprehensive dataset providing 11,890 diverse thermal queries referenced against a large-scale satellite ortho-photo and corresponding spatially aligned Digital Surface Model (DSM). Extensive experiments demonstrate that SCC-Loc establishes a new state-of-the-art, suppressing the mean localization error to 9.37 m and providing a 7.6-fold accuracy improvement within a strict 5-m threshold over the strongest baseline. Code and dataset are available at https://github.com/FloralHercules/SCC-Loc.

Abstract

Cross-View Geo-Localization

Explore similar work

SCC-Loc: A Unified Semantic Cascade Consensus Framework for UAV Thermal Geo-Localization

ScalLoc: Continuous Oblique Thermal UAV Geo-Localization with a 2D Satellite Map

Unifying UAV Cross-View Geo-Localization via 3D Geometric Perception