SDUNet: Road extraction via spatial enhanced and densely connected UNet

• Published in: Pattern recognition Vol. 126; p. 108549
• Main Authors: Yang, Mengxing, Yuan, Yuan, Liu, Ganchao
• Format: Journal Article
• Language: English
• Published: Elsevier Ltd 01.06.2022
• Subjects: Image segmentation; Remote sensing imagery; Road extraction; Spatial topology; Image segmentation; Road extraction; Spatial topology; Remote sensing imagery
• ISSN: 0031-3203; 1873-5142
• DOI: 10.1016/j.patcog.2022.108549

•We propose a network SDUNet, which combines the multi-level features of the road network and global prior information.•A structure preserving model is proposed to enhance feature learning about the structure prior of the road surface.•Experimental results show that our approach achieves state-of-the-art performance compared with previous methods. [Display omitted] Extracting road maps from high-resolution optical remote sensing images has received much attention recently, especially with the rapid development of deep learning methods. However, most of these CNN based approaches simply focused on multi-scale encoder architectures or multiple branches in neural networks, and ignored some inherent characteristics of the road surface. In this paper, we design a novel network for road extraction based on spatial enhanced and densely connected UNet, called SDUNet. SDUNet aggregates both the multi-level features and global prior information of road networks by combining the strengths of spatial CNN-based segmentation and densely connected blocks. To enhance the feature learning about prior information of road surface, a structure preserving model is designed to explore the continuous clues in the spatial level. Experimental results on two benchmark datasets show that the proposed method achieves the state-of-the-art performance, compared with previous approaches for road extraction. Code will be made available on https://github.com/MrStrangerYang/SDUNet.