The efficient extraction model of typical Chinese urban buildings is one of the important research contents to provide technical support for urban planning. However, the different architectural styles, a large number of vegetation and shadows in typical Chinese cities make the extraction of buildings with high errors. To address the problems of slow convergence speed and rough edge segmentation of the K-net combined with DeepLabv3 model on the remote sensing images of typical urban buildings in China, the Swin_ASPP_Knet network is proposed, which adds a multi-scale feature fusion module to enrich the feature information, and uses Swin Transformer to replace Resnet as the backbone network to extract more accurate feature information and accelerate the model convergence speed. After the comparative analysis of the experiments, the results show that the Swin_ASPP_Knet network outperforms K-Net combined with DeepLabv3 and Unet networks in the extraction task of typical urban building datasets in China, with mIoU and PA reaching 86.38% and 95.27%, respectively. To verify the generalization ability of the model, experiments were also conducted in the WHU building dataset where mIoU and PA reached 93.65% and 98.00%, respectively. Compared with the original model, the Swin_ASPP_Knet in both public building datasets reduced the model training time by half, and achieved good results in edge extraction of Chinese urban buildings with complex backgrounds, which met the requirement of efficient and accurate extraction of typical urban buildings in China.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.