HXHao Xu

Hao Xu

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Hao Xu has a total of 9 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    A Topology-Driven Cross-Modal Fusion Network for Winter Wheat Parcel Extraction From Optical and SAR Imagery

    by Zeren Li, Chenyang Liu, Yuanyuan Wang, Xiaoyu Zhang, Zhaoyun Cui, Huake Zhang, Hao Xu, Yanhui Jin, Chengming Zhang, Xiaoxia Yang

    Fine-scale winter wheat parcel extraction is critical for agricultural monitoring, planting area estimation, and spatial distribution analysis. However, existing methods are constrained by the limited representational capacity of single-source data and fail to fully exploit the complementary information provided by optical and synthetic aperture radar remote sensing data. To address this issue, this study proposes a Topology-Driven Cross-Modal Fusion Network for winter wheat parcel extraction. Specifically, a Topology-Prior State Space Block is introduced into the optical branch to enhance the modeling of regional connectivity, overall geometric structure, and long-range spatial dependencies of parcels. A Sparse Cross-Modal Guidance Block is further designed to improve the reliability of cross-modal interaction under the guidance of optical structural information, thereby reducing the adverse effects of synthetic aperture radar noise interference and false matching. In addition, a Spatial-Channel Gated Fusion Block is designed to enable adaptive fusion of multimodal features in terms of spatial responses, semantic channels, and modality contributions, thereby enhancing boundary delineation and result stability. A winter wheat dataset was constructed from Gaofen-6 optical imagery and Sentinel-1 synthetic aperture radar imagery covering Tai'an City and Lanling County, Linyi City, Shandong Province, and comparative and ablation experiments were performed. The proposed method achieves an F1-score of 0.887 and an intersection over union of 0.797, demonstrating its effectiveness and potential applicability in complex agricultural scenes and crop mapping tasks.