Topological Building Extraction With Bidirectional Prediction From Remote Sensing Images
Mingming Zhang,Ye Du,Zhenghui Hu,Wei Wang,Qingjie Liu,Yunhong Wang
DOI: https://doi.org/10.1109/jstars.2024.3399251
IF: 4.715
2024-05-31
IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
Abstract:Topological building extraction in remote sensing images is vital for city planning, disaster assessment, and other real-world applications. To meet the requirements of real-world applications, existing building extraction approaches predict topological building by vectorization of binary building masks using multiple refinement stages, leading to complex methodology and poor generalization. To tackle this issue, we propose a topological building extraction approach by directly predicting serialized vertices of each building instance. We observe that the order of serialized vertices from one building is inherently bidirectional, which can be clockwise or counterclockwise. By this new observation, the proposed method learns serialized vertices for each building supervised by the bidirectional constraint. Moreover, we design a cross-scale feature fusion module to obtain building representations with rich spatial and context information, facilitating the following serialized vertex prediction. Besides, a merge strategy is adopted to generate the final topological building from serialized vertices of two directions (clockwise and counterclockwise). Experiments are conducted on three building benchmarks to evaluate the effectiveness of our proposed method. Finally, extensive results show that the proposed approach outperforms state-of-the-art methods highlighting its superiority.
imaging science & photographic technology,remote sensing,engineering, electrical & electronic,geography, physical