geospatial data fusion

Density-equalizing map projections often suffer from topological failures—such as region disconnection or overlap—due to sparse boundary polygon vertices. To address this, we propose a conformal polyline densification method that, for the first time, rigorously guarantees regional connectivity and non-overlap in density-equalizing cartogram generation. Our approach integrates a flow-field-driven deformation framework, an adaptive boundary polyline subdivision strategy, and a geometry-topology co-verification mechanism, thereby preserving shape fidelity while enhancing structural consistency. Experimental evaluation demonstrates that our method outperforms state-of-the-art techniques in cartographic accuracy, computational efficiency, and topological robustness. It is particularly suitable for high-precision geographic data visualization, enabling reliable and interpretable spatial representations under significant density distortion.

Map-to-map matching faces three major challenges: absence of ground-truth correspondences, sparse node features, and poor scalability to large-scale maps. To address these, we propose the first fully unsupervised graph neural network framework. Our method introduces (1) pseudo-coordinate encoding to enrich node geometric representations and enable scale-invariant feature learning; (2) an adaptive feature-geometry similarity fusion mechanism jointly optimized with a geometric consistency loss to enhance robustness; and (3) a tiling-based overlapping partitioning strategy coupled with majority-voting post-processing to enable efficient parallel inference. Evaluated on real-world multi-source map datasets, our approach significantly outperforms existing supervised and unsupervised methods—achieving state-of-the-art accuracy, especially under high noise and at large scale. The results demonstrate its effectiveness, scalability, and practical utility for real-world map alignment tasks.

This paper identifies and systematically analyzes common challenges in spatial data science across mainstream programming languages—R, Python, and Julia—including inconsistent spherical geometry modeling, ambiguous spatial/temporal semantics, conflation of intensive and extensive attributes, poor interoperability between data cube and vector formats, complex cross-package dependencies, and a persistent divide between GIS and physical modeling communities. Through multi-language ecosystem surveys, cross-community comparative analysis, and software engineering abstraction, we propose, for the first time, a cross-language semantic framework for spatial operations. The framework formally defines support types (point vs. block), specifies attribute-type constraints on operation validity, and refactors spherical Simple Features logic. We distill five foundational insights that establish a methodological basis and practical guidance for tool interoperability, pedagogical alignment, and open-source governance in spatial computing.

Existing building vectorization methods heavily rely on image modalities and suffer from insufficient geometric accuracy. To address this, we introduce the first intercontinental, multimodal building vectorization benchmark, integrating 25 cm GSD aerial imagery, decimeter-level LiDAR point clouds (>1 billion points), and high-precision ground-truth vector footprints. We conduct the first systematic evaluation of LiDAR’s robustness for building vectorization and propose novel multimodal alignment strategies, joint point cloud–image encoding, and both end-to-end and two-stage prediction frameworks based on Transformers and CNNs. We publicly release the dataset, training code, and weights of three state-of-the-art models. Experiments demonstrate that image–LiDAR fusion substantially outperforms unimodal approaches, improving F1 and IoU by 8–12% on average, while significantly enhancing polygonal geometric integrity and topological correctness.

This work addresses the limitations of existing Earth observation foundation models, which predominantly rely on raster data and overlook the structured geographic semantics embedded in open vector datasets such as OpenStreetMap, thereby hindering comprehensive understanding of human–environment systems. To overcome this, we propose the first unified spatial representation learning framework that deeply integrates remote sensing imagery and vector data within a shared embedding space, breaking away from conventional modality-isolated paradigms. By leveraging self-supervised learning and multimodal alignment—while explicitly modeling geometric, topological, and semantic relationships—our approach enables synergistic raster perception and vector-based reasoning. The method substantially enhances accuracy, semantic interpretability, and explainability on downstream tasks, laying a theoretical and methodological foundation for developing human-centered, semantically rich geospatial foundation models.

This work addresses the communication bottleneck in multi-robot LiDAR-based mapping, where massive sensor data transmission hinders efficient collaboration, particularly under resource-constrained conditions. The study introduces a novel framework that formulates map merging as a three-stage cascaded optimization problem over an exchange graph. By leveraging graph-theoretic methods to select a critical subset of scans, the approach enables lightweight data exchange through joint geometric and perceptual optimization. The proposed method requires transmitting only a small number of keyframes, achieving significant reductions in communication overhead while preserving alignment accuracy. Experiments on multiple public and self-collected datasets demonstrate up to a 99.98% reduction in communication volume—e.g., from 7000 MB to 1.3 MB—and confirm its applicability across platforms ranging from embedded systems to desktop computers.

Optical imagery-based building footprint extraction is often compromised by occlusion, perspective distortion, and the absence of elevation data, leading to incomplete or misaligned results. This work proposes the first large-scale vectorized building footprint dataset and benchmark specifically designed for airborne LiDAR point clouds, encompassing 33,000 urban and rural tiles of size 128×128 meters, along with 3,000 cross-domain test tiles to evaluate geographic generalization. The dataset provides precisely aligned vector footprints coupled with elevation information, enabling fine-grained modeling and cross-regional studies. Through comprehensive baseline experiments, this study highlights key challenges including high intra-class variability, data imbalance, and noise, thereby advancing research in urban perception and building modeling.

Existing spatial pattern matching methods are largely confined to two-dimensional space and struggle to handle three-dimensional entity matching involving elevation or height information in real-world scenarios. This work extends spatial pattern matching to 3D environments for the first time, introducing a general problem formulation and proposing a subgraph-matching-based algorithm that explicitly models distance relationships in three-dimensional space. To support empirical evaluation, the authors construct the first 3D spatial pattern matching dataset, integrating both synthetic data and real-world building structures from the city of Hamburg. Experimental results on this benchmark demonstrate the effectiveness of the proposed approach, establishing a foundational algorithmic framework and experimental platform for future research in 3D spatial pattern analysis.

This work addresses the challenge of unified modeling for multi-class heterogeneous geographic entities in remote sensing vector mapping, where existing methods struggle to adequately represent topological relationships and instance boundaries. The authors propose reframing vector map construction as a structured text generation task by designing a GeoJSON-like hierarchical vector language that jointly encodes geometric, semantic, and topological information. A progressive vision-to-language mapping framework is introduced, optimized via reinforcement learning to ensure syntactic validity, content fidelity, and map executability of the generated output, thereby enabling cross-category unified modeling. Experiments on the newly curated VecMap-Bench dataset—comprising 54K images and 800K instances—demonstrate that the proposed approach significantly outperforms state-of-the-art methods in single- and multi-class mapping, cross-dataset transfer, and open-vocabulary generalization.