Remote Sensing Image Scene Classification Method Integrating Spatial Transformation Structure and Depth Residual Network

被引：0
|
作者：
Meng Y. [1 ]
Zheng G. [1 ]
Ji W. [2 ]
机构：
[1] School of Geography and Information Engineering, China University of Geosciences, Wuhan
[2] School of Architectural and Surveying and Mapping Engineering, Jiangxi University of Science and Technology, Ganzhou
来源：
Diqiu Kexue - Zhongguo Dizhi Daxue Xuebao/Earth Science - Journal of China University of Geosciences | 2023年 / 48卷 / 09期
关键词：
deep learning; remote sensing; residual network; scene classification; spatial transformation networks; transfer learning;
D O I：
10.3799/dqkx.2021.218
中图分类号：
学科分类号：
摘要：
In order to solve the problem that the remote sensing image with small sample set can easily lead to the over-fitting of the training model and the low classification accuracy caused by the spatial invariance of convolution neural network in remote sensing image scene classification, a high-resolution remote sensing image scene classification algorithm based on spatial transformation network and transfer learning is proposed. Firstly, the ImageNet dataset is used to train the deep residual network ResNet101 to obtain the pre-training model, and the training efficiency of the model is improved through knowledge transfer. Then, the spatial transformation structure is embedded in the model, so that the model can actively transform the feature mapping in space and improve the robustness of the model. Finally, the Dropout layer is added to the model to reduce the probability of over-fitting of the model. This method is verified on two high-score remote sensing image data sets of AID and NWPURESISC45, and the classification accuracy of 94.30% and 93.63% is achieved in the case of only 20% training samples. The experimental results show that the improved model has better feature extraction ability and better classification results for misclassification scenarios. © 2023 China University of Geosciences. All rights reserved.
引用
收藏
页码：3526 / 3538
页数：12
相关论文
共 37 条
  • [1] Berman M., Triki A. R., Blaschko M. B., The Lo⁃ vasz⁃Softmax Loss: A Tractable Surrogate for the Opti⁃ mization of the Intersection⁃Over⁃Union Measure in Neu⁃ ral Networks, 2018 IEEE/CVF Conference on Comput⁃ er Vision and Pattern Recognition, (2018)
  • [2] Cheng G., Guo L., Zhao T. Y., Et al., Auto⁃ matic Landslide Detection from Remote⁃Sensing Imag⁃ ery Using a Scene Classification Method Based on BoVW and pLSA, International Journal of Remote Sensing, 34, 1, pp. 45-59, (2013)
  • [3] Cheng G., Han J. W., Lu X. Q., Remote Sensing Image Scene Classification: Benchmark and State of the Art, Proceedings of the IEEE, 105, 10, pp. 1865-1883, (2017)
  • [4] Cheng G., Yang C. Y., Yao X. W., Et al., When Deep Learning Meets Metric Learning: Remote Sensing Image Scene Classification via Learning Discriminative CNNS, IEEE Transactions on Geoscience and Remote Sensing, 56, 5, pp. 2811-2821, (2018)
  • [5] Cheng G. X., Niu R. Q., Zhang K. X., Et al., Opencast Mining Area Recognition in High ⁃ Resolution Remote Sensing Images Using Convolutional Neural Networks, Earth Science, 43, pp. 256-262, (2018)
  • [6] Dalal N., Triggs B., Histograms of Oriented Gradi⁃ ents for Human Detection, 2005 IEEE Computer Soci⁃ ety Conference on Computer Vision and Pattern Recog⁃ nition (CVPR'05), (2005)
  • [7] Donahue J., Jia Y. Q., Vinyals O., Et al., DeCAF: A Deep Convolutional Activation Feature for Generic Visual Recognition, Proceedings of the 31st Internation⁃ al Conference on Machine Learning, (2014)
  • [8] Han X., Zhang Z. Y., Ding N., Et al., Pre⁃Trained Models: Past, Present and Future, AI Open, 2, pp. 225-250, (2021)
  • [9] He K. M., Zhang X. Y., Ren S. Q., Et al., Deep Residual Learning for Image Recognition, 2016 IEEE Conference on Computer Vision and Pattern Recognition, (2016)
  • [10] He N. J., Fang L. Y., Li S. T., Et al., Remote Sensing Scene Classification Using Multilayer Stacked Covariance Pooling, IEEE Transactions on Geoscience and Remote Sensing, 56, 12, pp. 6899-6910, (2018)
1234