Boundary-Aware Multiscale Learning Perception for Remote Sensing Image Segmentation

For remote sensing image segmentation, the boundaries of objects are difficult to distinguish, which is ignored by most methods. Therefore, it is challenging how to excavate and recover the boundaries of objects accurately. In this article, we propose a boundary-aware multiscale network (BMNet) to solve this problem. The key components of BMNet include the scale attention module (SAM) and boundary guidance module (BGM). Specifically, the SAM is proposed to guide the refinement of multiscale features in a context-aware way. It enhances the discriminability of multiscale features by establishing contextual dependencies, which enables the refinement of the prediction of objects. Then, the BGM is proposed to enable networks to distinguish the boundary of objects. It uses manifold information of features to generate BG maps and forces the network to focus more on the boundary of objects. The effectiveness of the proposed BMNet is demonstrated on two public remote sensing datasets: ISPRS 2-D semantic labeling Potsdam dataset and Vaihingen dataset, where BMNet achieves better segmentation than prevalent methods. Finally, the experimental results indicate that BMNet can produce sharper boundaries of objects to reconstruct more detailed segmentation results.