Semantic segmentation of remote sensing images based on U-shaped network combined with spatial enhance attention

被引：0
|
作者：
Bao Y. [1 ,2 ]
Liu W. [1 ]
Li R. [1 ]
Li Q. [1 ]
Hu Q. [1 ]
机构：
[1] School of Data and Target Engineering, PLA Strategic Support Force Information Engineering University, Zhengzhou
[2] Unit 31401 of PLA, Hohhot
来源：
Beijing Hangkong Hangtian Daxue Xuebao/Journal of Beijing University of Aeronautics and Astronautics | 2023年 / 49卷 / 07期
基金：
中国国家自然科学基金;
关键词：
attention; deep learning; loss function; remote sensing image; semantics segmentation;
D O I：
10.13700/j.bh.1001-5965.2021.0544
中图分类号：
学科分类号：
摘要：
The performance of semantic segmentation based on deep learning still need to be improved when analyzing small-sized objects and object boundaries in remote sensing images. Aiming at this problem, we propose a U-shaped network (SGE-Unet). Firstly, the structure of the model is optimized to enhance the representation of feature. Secondly, we add the attention module of spatial group enhance to extract semantic information. Finally, the median frequency balance cross-entropy loss function is used to suppress the unbalanced distribution of classes. The experiment was conducted on two datasets and shows that the overall accuracy,mean interaction over union, F1, and Kappa of SGE-Unet are better than mainstream models. In experiments of the Vaihingen dataset, the interaction over union and F1 of the car reached 0.719 and 0.901, which were 16% and 11% higher than those of the model with the second-highest performance. The experimental results show that the proposed module greatly improves the segmentation of easily confused objects, small-sized objects, and object boundaries. © 2023 Beijing University of Aeronautics and Astronautics (BUAA). All rights reserved.
引用
收藏
页码：1828 / 1837
页数：9
相关论文
共 27 条
  • [1] YUAN X H, SHI J F, GU L C., A review of deep learning methods for semantic segmentation of remote sensing imagery, Expert Systems with Applications, 169, (2021)
  • [2] XING S, XIE Q, WANG M., Semantic segmentation for remote sensing images based on adaptive feature selection network, IEEE Geoscience and Remote Sensing Letters, 19, (2022)
  • [3] JIANG C C, HUO H T, FENG Q., An object-oriented multi-scale segmentation optimization algorithm based on PCA, Journal of Beijing University of Aeronautics and Astronautics, 46, 6, pp. 1192-1203, (2020)
  • [4] KAMPFFMEYER M, SALBERG A B, JENSSEN R., Semantic segmentation of small objects and modeling of uncertainty in urban remote sensing images using deep convolutional neural networks, IEEE Conference on Computer Vision and Pattern Recognition Workshops (CVPRW), pp. 680-688, (2016)
  • [5] SHELHAMER E, LONG J, DARRELL T., Fully convolutional networks for semantic segmentation, IEEE Transactions on Pattern Analysis and Machine Intelligence, 39, 4, pp. 640-651, (2017)
  • [6] GUO R, LIU J B, LI N, Et al., Pixel-wise classification method for high resolution remote sensing imagery using deep neural networks, ISPRS International Journal of Geo-Information, 7, 3, (2018)
  • [7] LI R, DUAN C X, ZHENG S Y, Et al., MACU-Net for semantic segmentation of fine-resolution remotely sensed images, IEEE Geoscience and Remote Sensing Letters, 19, (2022)
  • [8] ALAM M, WANG J F, CONG G P, Et al., Convolutional neural network for the semantic segmentation of remote sensing images, Mobile Networks and Applications, 26, 1, pp. 200-215, (2021)
  • [9] RONNEBERGER O, FISCHER P, BROX T., U-Net: Convolutional networks for biomedical image segmentation, International Conference on Medical Image Computing and Computer-Assisted Intervention, pp. 234-241, (2015)
  • [10] ZHANG X J, WANG X L., Image segmentation models of remote sensing using full residual connection and multiscale feature fusion, Journal of Remote Sensing, 24, 9, pp. 1120-1133, (2020)
123