Research article Special Issues

Multi-source remote sensing image classification based on two-channel densely connected convolutional networks

  • Received: 24 July 2020 Accepted: 09 October 2020 Published: 27 October 2020
  • Remote sensing image classification exploiting multiple sensors is a very challenging problem: The traditional methods based on the medium- or low-resolution remote sensing images always provide low accuracy and poor automation level because the potential of multi-source remote sensing data are not fully utilized and the low-level features are not effectively organized. The recent method based on deep learning can efficiently improve the classification accuracy, but as the depth of deep neural network increases, the network is prone to be overfitting. In order to address these problems, a novel Two-channel Densely Connected Convolutional Networks (TDCC) is proposed to automatically classify the ground surfaces based on deep learning and multi-source remote sensing data. The main contributions of this paper includes the following aspects: First, the multi-source remote sensing data consisting of hyperspectral image (HSI) and Light Detection and Ranging (LiDAR) are pre-processed and re-sampled, and then the hyperspectral data and LiDAR data are input into the feature extraction channel, respectively. Secondly, two-channel densely connected convolutional networks for feature extraction were proposed to automatically extract the spatial-spectral feature of HSI and LiDAR. Thirdly, a feature fusion network is designed to fuse the hyperspectral image features and LiDAR features. The fused features were classified and the output result is the category of the corresponding pixel. The experiments were conducted on popular dataset, the results demonstrate that the competitive performance of the TDCC with respect to classification performance compared with other state-of-the-art classification methods in terms of the OA, AA and Kappa, and it is more suitable for the classification of complex ground surfaces.

    Citation: Haifeng Song, Weiwei Yang, Songsong Dai, Haiyan Yuan. Multi-source remote sensing image classification based on two-channel densely connected convolutional networks[J]. Mathematical Biosciences and Engineering, 2020, 17(6): 7353-7377. doi: 10.3934/mbe.2020376

    Related Papers:

  • Remote sensing image classification exploiting multiple sensors is a very challenging problem: The traditional methods based on the medium- or low-resolution remote sensing images always provide low accuracy and poor automation level because the potential of multi-source remote sensing data are not fully utilized and the low-level features are not effectively organized. The recent method based on deep learning can efficiently improve the classification accuracy, but as the depth of deep neural network increases, the network is prone to be overfitting. In order to address these problems, a novel Two-channel Densely Connected Convolutional Networks (TDCC) is proposed to automatically classify the ground surfaces based on deep learning and multi-source remote sensing data. The main contributions of this paper includes the following aspects: First, the multi-source remote sensing data consisting of hyperspectral image (HSI) and Light Detection and Ranging (LiDAR) are pre-processed and re-sampled, and then the hyperspectral data and LiDAR data are input into the feature extraction channel, respectively. Secondly, two-channel densely connected convolutional networks for feature extraction were proposed to automatically extract the spatial-spectral feature of HSI and LiDAR. Thirdly, a feature fusion network is designed to fuse the hyperspectral image features and LiDAR features. The fused features were classified and the output result is the category of the corresponding pixel. The experiments were conducted on popular dataset, the results demonstrate that the competitive performance of the TDCC with respect to classification performance compared with other state-of-the-art classification methods in terms of the OA, AA and Kappa, and it is more suitable for the classification of complex ground surfaces.


    加载中


    [1] X. Yang, Y. Ye, X. Li, R. Y. K. Lau, X. Zhang, X. Huang, Hyperspectral image classification with deep learning models, IEEE Trans. Geosci. Remote Sens., 56 (2018), 5408-5423. doi: 10.1109/TGRS.2018.2815613
    [2] J. A. Benediktsson, I. Kanellopoulos, Classification of multisource and hyperspectral data based on decision fusion, IEEE Trans. Geosci. Remote Sens., 37 (1999), 1367-1377. doi: 10.1109/36.763301
    [3] B. Chen, B. Huang, B. Xu, Multi-source remotely sensed data fusion for improving land cover classification, Isprs J. Photogramm. Remote Sens., 124 (2017), 27-39. doi: 10.1016/j.isprsjprs.2016.12.008
    [4] Z. Mahmood, M. A. Akhter, G. Thoonen, P. Scheunders, Contextual subpixel mapping of hyperspectral images making use of a high resolution color image. IEEE J. Sel. Top. Appl. Earth Obs. Remote Sens., 6 (2013), 779-791. doi: 10.1109/JSTARS.2012.2236539
    [5] D. G. Goodenough, A. Dyk, K. O. Niemann, J. S. Pearlman, H. Chen, T. Han, et al., Processing hyperion and ali for forest classification, IEEE Trans. Geosci. Remote Sens., 41 (2003), 1321- 1331. doi: 10.1109/TGRS.2003.813214
    [6] D. G. Stavrakoudis, E. Dragozi, I. Z. Gitas, C. Karydas, Decision fusion based on hyperspectral and multispectral satellite imagery for accurate forest species mapping, Remote Sens., 6 (2014), 6897-6928. doi: 10.3390/rs6086897
    [7] T. Kattenborn, J. Maack, F. E. Fassnacht, F. Enssle, Corrigendum to mapping forest biomass from space fusion of hyperspectraleo1-hyperion data and tandem-x and worldview-2 canopy heightmodels [Int. J. Appl. Earth Obs. Geoinf. Issue no. 35 (2015) 359-367]. Int. J. Appl. Earth Obs. Geoinf., 41 (2014).
    [8] S. Delalieux, P. J. Zarco-Tejada, L. Tits, M. A. Jimenez Bello, D. Intrigliolo, B. Somers, Unmixing-based fusion of hyperspatial and hyperspectral airborne imagery for early detection of vegetation stress, IEEE J. Sel. Top. Appl. Earth Obs. Remote Sens., 7 (2014), 2571-2582. doi: 10.1109/JSTARS.2014.2330352
    [9] C. D. Packard, T. S. Viola, M. D. Klein. Hyperspectral target detection analysis of a cluttered scene from a virtual airborne sensor platform using muses, Proceedings of Target and Background Signatures, 2017.
    [10] J. R. Kaufman, M. T. Eismann, M. Celenk, Assessment of spatialspectral feature-level fusion for hyperspectral target detection, IEEE J. Sel. Top. Appl. Earth Obs. Remote Sens., 8 (2015), 2534-2544. doi: 10.1109/JSTARS.2015.2420651
    [11] N. B. Chang, B. Vannah, Y. J. Yang, Comparative sensor fusion between hyperspectral and multispectral satellite sensors for monitoring microcystin distribution in lake erie, IEEE J. Sel. Top. Appl. Earth Obs. Remote Sens., 7 (2014), 2426-2442. doi: 10.1109/JSTARS.2014.2329913
    [12] M. Dalponte, L. Bruzzone, D. Gianelle, Fusion of hyperspectral and lidar remote sensing data for classification of complex forest areas, IEEE Trans. Geosci. Remote Sens., 46 (2008), 1416-1427. doi: 10.1109/TGRS.2008.916480
    [13] A. Merentitis, C. Debes, R. Heremans, Ensemble learning in hyperspectral image classification: Toward selecting a favorable bias-variance tradeoff. IEEE J. Sel. Top. Appl. Earth Obs. Remote Sens., 7 (2014), 1089-1102. doi: 10.1109/JSTARS.2013.2295513
    [14] C. Debes, A. Merentitis, R. Heremans, J. Hahn, N. Frangiadakis, T. van Kasteren, et al., Hyperspectral and lidar data fusion: Outcome of the 2013 grss data fusion contest, IEEE J. Sel. Top. Appl. Earth Obs. Remote Sens., 7 (2014), 2405-2418. doi: 10.1109/JSTARS.2014.2305441
    [15] C. Chen, X. Fan, C. Zheng, L. Xiao, M. Cheng, C. Wang, Sdcae: Stack denoising convolutional autoencoder model for acc, 2018 Sixth International Conference on Advanced Cloud and Big Data (CBD), 2018.
    [16] A. Krizhevsky, I. Sutskever, G. Hinton, Imagenet classification with deep convolutional neural networks, Advances in neural information processing systems, 2012.
    [17] Y. Chen, Z. Lin, X. Zhao, G. Wang, Y. Gu, Deep learning-based classification of hyperspectral data, IEEE J. Sel. Top. Appl. Earth Obs. Remote Sens., 7 (2014), 2094-2107. doi: 10.1109/JSTARS.2014.2329330
    [18] X. Ma, H. Wang, J. Geng, Spectralspatial classification of hyperspectral image based on deep auto-encoder, IEEE J. Sel. Top. Appl. Earth Obs. Remote Sens., 9 (2016), 4073-4085. doi: 10.1109/JSTARS.2016.2517204
    [19] A. Mughees, L. Tao. Efficient deep auto-encoder learning for the classification of hyperspectral images, In 2016 International Conference on Virtual Reality and Visualization (ICVRV), 2016.
    [20] J. Leng, T. Li, G. Bai, Q. Dong, D. Han. Cube-cnn-svm: A novel hyperspectral image classification method, In 2016 IEEE 28th International Conference on Tools with Artificial Intelligence (ICTAI), 2016.
    [21] Y. Li, H. Zhang, Q. Shen, Spectralspatial classification of hyperspectral imagery with 3d convolutional neural network, Remote Sens., 9 (2017), 67. doi: 10.3390/rs9010067
    [22] J. Yue, W. Zhao, S. Mao, and H. Liu, Spectral-spatial classification of hyperspectral images using deep convolutional neural networks. Remote Sens. Lett., 6 (2015), 468-477. doi: 10.1080/2150704X.2015.1047045
    [23] W. Zhao, S. Du. Spectralspatial feature extraction for hyperspectral image classification: A dimension reduction and deep learning approach, IEEE Trans. Geosci. Remote Sens., 54 (2016), 4544-4554. doi: 10.1109/TGRS.2016.2543748
    [24] A. Santara, K. Mani, P. Hatwar, A. Singh, A. Garg, K. Padia, et al., Bass net: Band-adaptive spectral-spatial feature learning neural network for hyperspectral image classification, IEEE Trans. Geosci. Remote Sens., 55 (2017), 5293-5301. doi: 10.1109/TGRS.2017.2705073
    [25] Y. Chen, H. Jiang, C. Li, X. Jia, P. Ghamisi, Deep feature extraction and classification of hyperspectral images based on convolutional neural networks, IEEE Trans. Geosci. Remote Sens., 54 (2016), 6232-6251. doi: 10.1109/TGRS.2016.2584107
    [26] S. Wu, S. Zhong, Y. Liu, Deep residual learning for image steganalysis, Multimedia Tools Appl., 77 (2018), 10437-10453. doi: 10.1007/s11042-017-4476-5
    [27] X. Glorot, Y. Bengio, Understanding the difficulty of training deep feedforward neural networks, Proceedings of the Thirteenth International Conference on Artificial Intelligence and Statistics, 2010.
    [28] Y. Lecun, L. Bottou, Y. Bengio, P. Haffner, Gradient-based learning applied to document recognition, Proc. IEEE, 86 (1998), 2278-2324. doi: 10.1109/5.726791
    [29] K. Simonyan, A. Zisserman, Very deep convolutional networks for large-scale image recognition. Comput. Sci., 2015 (2015), 1-14.
    [30] G. Huang, Z. Liu, L. van der Maaten, K. Q. Weinberger, Densely connected convolutional networks, 2018 IEEE Conference on Computer Vision and Pattern Recognition, 2017.
    [31] S. Ioffe, C. Szegedy, Batch normalization: Accelerating deep network training by reducing internal covariate shift, Proceedings of the 32nd International Conference on Machine Learning, 2015.
    [32] G. E. Hinton, N. Srivastava, A. Krizhevsky, I. Sutskever, R. R. Salakhutdinov, Improving neural networks by preventing co-adaptation of feature detectors, Comput. Sci., 2012 (2012), 212-223.
    [33] W. Hu, Y. Huang, L. Wei, F. Zhang, H. Li, Deep convolutional neural networks for hyperspectral image classification, J. Sensors, 2015(2015):112, 2015.
    [34] W. Jing, S. Huo, Q. Miao, X. Chen, A model of parallel mosaicking for massive remote sensing images based on spark, IEEE Access, 5 (2017), 18229-18237. doi: 10.1109/ACCESS.2017.2746098
  • Reader Comments
  • © 2020 the Author(s), licensee AIMS Press. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0)
通讯作者: 陈斌, bchen63@163.com
  • 1. 

    沈阳化工大学材料科学与工程学院 沈阳 110142

  1. 本站搜索
  2. 百度学术搜索
  3. 万方数据库搜索
  4. CNKI搜索

Metrics

Article views(3284) PDF downloads(154) Cited by(3)

Article outline

Figures and Tables

Figures(15)  /  Tables(5)

Other Articles By Authors

/

DownLoad:  Full-Size Img  PowerPoint
Return
Return

Catalog