Back

Deep Learning for Hyperspectral Image Classification: An Overview

Shutao LiWeiwei SongLeyuan FangYushi ChenPedram GhamisiJón Atli Benediktsson

article2019IEEE Transactions on Geoscience and Remote Sensing1,628 citationsIEEE GRSS 2020 Highest Impact Paper Award

Presents a systematic framework categorizing deep learning approaches for hyperspectral image classification into spectral, spatial, and joint spectral-spatial networks while evaluating practical strategies to overcome limited training data constraints in remote sensing.

Listen

Hyperspectral imaging collects data across hundreds of narrow spectral bands to identify surface materials on Earth. However, accurately classifying this data is difficult due to complex nonlinear relationships, high environmental variability, and the challenge of having very high-dimensional data paired with scarce ground-truth labels. Traditional machine learning methods rely heavily on hand-crafted features and expert tuning, which struggle to generalize across diverse operational settings. The article addresses this challenge by examining how modern deep learning techniques can automatically extract rich, hierarchical features to improve land-cover and material classification.

The main objective of the article is to provide a structured overview of deep learning methodologies for hyperspectral image classification, assess strategies for training models when labeled data is scarce, and benchmark deep learning models against traditional algorithms.

To accomplish this, the article surveys the research landscape by categorizing methods into spectral, spatial, and joint spectral-spatial feature extraction frameworks. It then conducts controlled empirical evaluations using three standard hyperspectral benchmark datasets: Houston, University of Pavia, and Salinas. The experiments test classical baselines such as Support Vector Machines against advanced deep learning architectures, while also isolating the impact of practical strategies such as data augmentation, transfer learning, and residual connections under constrained training conditions.

The analysis reveals several key findings. First, deep learning models consistently outperform traditional machine learning classifiers across all benchmarks, generating cleaner maps with substantially fewer noisy misclassifications. Second, joint spectral-spatial networks achieve the highest overall performance; for example, the Deep Feature Fusion Network delivered top-tier overall accuracy on all datasets, reaching 84.56% on Houston, 98.57% on University of Pavia, and 99.71% on Salinas. Third, integrating structural spatial filters into deep architectures significantly boosts accuracy over spatial filtering alone, as demonstrated by a Gabor-filtered convolutional network achieving an overall accuracy approximately 3.5 percentage points higher than standard edge-preserving filtering on the Houston data. Finally, when labeled training samples are extremely limited, residual network optimization delivers the most reliable performance gains, while transfer learning can improve accuracy by roughly 1% when using as few as five labeled samples per class.

These findings indicate that transitioning from manual feature engineering to deep hierarchical modeling enhances mapping accuracy and operational reliability in remote sensing. By capturing subtle spatial-contextual and spectral patterns simultaneously, deep architectures reduce false alarms and inconsistent boundaries. Furthermore, the demonstrated success of transfer learning and residual optimization means organizations can deploy high-performing models even when obtaining extensive field-collected training samples is cost-prohibitive.

For future implementations, decision-makers and technical teams should adopt integrated spectral-spatial architectures, specifically leveraging residual learning and transfer learning when working with small labeled datasets. When configuring deep networks, practitioners should tailor architectures to the physical characteristics of hyperspectral cubes rather than relying purely on off-the-shelf computer vision designs. While the evidence supporting these methods is robust across the evaluated benchmarks, users should maintain appropriate caution regarding high computational costs and the current reliance on trial-and-error network tuning for feature fusion.

Harbin Institute of TechnologyHelmholtz Institute Freiberg for Resource TechnologyHelmholtz-Zentrum Dresden–RossendorfHunan UniversityKey Laboratory of Visual Perception and Artificial Intelligence of Hunan ProvinceUniversity of Icelanddeep learningfeature extractionhyperspectral image classificationhyperspectral image classificationdeep learningspectral-spatial feature-based classification
arXiv: 1910.12861
  • Paper: Image Segmentation Using Deep Learning: A Survey, Shervin Minaee et al. (2020). This survey expands upon deep learning pixel-level classification by examining advanced encoder-decoder and semantic segmentation paradigms across broader computer vision domains.
  • Paper: ResUNet-a: a deep learning framework for semantic segmentation of remotely sensed data, Foivos I. Diakogiannis et al. (2019). This work develops an advanced multi-scale, conditioned segmentation framework for remote sensing that directly extends the deep spatial and spectral modeling principles discussed in the survey.
  • Paper: U2Fusion: A Unified Unsupervised Image Fusion Network, Han Xu et al. (2020). This paper advances multi-modal remote sensing and image representation by proposing a unified unsupervised network to fuse complementary image modalities without labeled ground truth.
  • Paper: Self-Supervised Learning: Generative or Contrastive, Xiao Liu et al. (2020). This survey deepens the strategies for overcoming limited training samples highlighted in the source by reviewing state-of-the-art generative and contrastive self-supervised representation learning.
  • Paper: Ensemble deep learning: A review, M. A. Ganaie et al. (2021). This review investigates deep ensemble techniques that generalize decision-fusion strategies for mitigating sample scarcity and variance in complex classification pipelines.
Cover for Deep Learning for Hyperspectral Image Classification: An Overview

Abstract

Hyperspectral image (HSI) classification has become a hot topic in the field of remote sensing. In general, the complex characteristics of hyperspectral data make the accurate classification of such data challenging for traditional machine learning methods. In addition, hyperspectral imaging often deals with an inherently nonlinear relation between the captured spectral information and the corresponding materials. In recent years, deep learning has been recognized as a powerful feature-extraction tool to effectively address nonlinear problems and widely used in a number of image processing tasks. Motivated by those successful applications, deep learning has also been introduced to classify HSIs and demonstrated good performance. This survey paper presents a systematic review of deep learning-based HSI classification literatures and compares several strategies for this topic. Specifically, we first summarize the main challenges of HSI classification which cannot be effectively overcome by traditional machine learning methods, and also introduce the advantages of deep learning to handle these problems. Then, we build a framework which divides the corresponding works into spectral-feature networks, spatial-feature networks, and spectral-spatial-feature networks to systematically review the recent achievements in deep learning-based HSI classification. In addition, considering the fact that available training samples in the remote sensing field are usually very limited and training deep networks require a large number of samples, we include some strategies to improve classification performance, which can provide some guidelines for future studies on this topic. Finally, several representative deep learning-based classification methods are conducted on real HSIs in our experiments.

Topics

deep learningfeature extractionhyperspectral image classificationtraining sampleclassification modelsconvolutional neural networksnetwork architecturesupervised classificationtransfer learning

Keywords

hyperspectral image classificationdeep learningspectral-spatial feature-based classificationdeep convolutional neural networkdeep feature fusion networkremote sensingresidual learningpixel-wise classifiers

Table of Contents

  • I Introduction
  • II Deep Models
  • II-A SAEs
  • II-B DBNs
  • II-C CNNs
  • II-C1 Convolutional Layers
  • II-C2 Pooling Layers
  • II-C3 Fully Connected Layers
  • II-D RNNs
  • II-E GANs
  • III Deep Networks-based HSI Classification
  • III-A Spectral-Feature Networks
  • III-B Spatial-Feature Networks
  • III-C Spectral-Spatial-Feature Networks
  • III-C1 Preprocessing-based networks
  • III-C2 Integrated networks
  • III-C3 Postprocessing-based networks
  • IV Strategies for Limited Available Samples
  • IV-A Data Augmentation
  • IV-A1 Transformation-based sample generation
  • IV-A2 Mixture-based sample generation
  • IV-B Transfer Learning
  • IV-C Unsupervised/Semi-supervised Feature Learning
  • IV-D Network Optimization
  • V Experiments
  • V-A Experimental Data Sets
  • V-B Compared Methods
  • V-C Classification Results
  • V-D Deep Feature Visualization
  • V-E Effectiveness Analysis of Strategies for Limited Samples
  • VI Conclusion
  • References

Knowls

  1. Knowl 1 — Taxonomy of Deep Learning Architectures for Hyperspectral Image Classification

    model/method

    Deep learning architectures for hyperspectral image (HSI) classification are categorized based on the feature modalities they extract:

    1. Spectral-Feature Networks: These networks process purely spectral vectors x∈Rdx \in \mathbb{R}^d across dd spectral bands to extract deep invariant representations without spatial context. Typical architectures include Stacked Auto-Encoders (SAEs), Deep Belief Networks (DBNs), 1-D Convolutional Neural Networks (1-D CNNs), and Recurrent Neural Networks (RNNs) that treat spectral channels as ordered sequence data.

    2. Spatial-Feature Networks: These frameworks apply spatial operations on dimensionally reduced hyperspectral cubes (e.g., using principal component analysis to project the spectral dimension down to principal components) using 2-D CNNs to extract texture and morphological spatial features from local pixel neighborhoods, which are subsequently concatenated or fused with spectral vectors.

    3. Spectral-Spatial-Feature Networks: These architectures extract joint spectral-spatial representations and are subdivided into three structural paradigms:

      • Preprocessing-Based Networks: Spatial and spectral features are fused at a low level prior to deep feature learning (e.g., spatial neighborhood flattening, spatial pixel averaging within local windows, or filtering via Gabor/attribute/rolling-guidance filters) before being passed to fully connected deep networks (SAE, DBN) or simple classifiers.
      • Integrated Networks: Joint spectral-spatial features are extracted end-to-end directly from 3-D input cubes using 3-D convolutions (3-D CNN), fully convolutional networks (FCNs), deep residual networks (DRNs), capsule networks, or hybrid convolutional-recurrent architectures (CRNNs).
      • Postprocessing-Based Networks: Spectral features and spatial features are extracted separately through two parallel deep network branches (e.g., a 1-D CNN branch for spectral features and a 2-D CNN branch for spatial patches, or two-stream autoencoder-CNN setups) and subsequently fused in fully connected layers for final classification.

  2. Knowl 2 — Data Augmentation Techniques for Limited Hyperspectral Training Samples

    equation

    To mitigate the scarcity of labeled training data in hyperspectral image classification, two primary formulations generate synthetic labeled samples from existing labeled pixels:

    1. Transformation-Based Sample Generation:

    y=f(xi)+γny = f(x_i) + \gamma n

    where xi∈Rdx_i \in \mathbb{R}^d is a known training sample across dd spectral bands, f(⋅)f(\cdot) is a transformation function (such as rotation, flipping, or mirroring of spatial neighborhoods), nn is a random Gaussian noise vector modeling sensor error and pixel interaction, γ\gamma is a scalar controlling the weight of the noise, and yy is the resulting virtual sample assigned the class label of xix_i.

    1. Mixture-Based Sample Generation:

    y=αijxi+(1−αij)xjy = \alpha_{ij} x_i + (1 - \alpha_{ij}) x_j

    where xi,xj∈Rdx_i, x_j \in \mathbb{R}^d are two known training samples belonging to the same land-cover class, and αij∈(0,1]\alpha_{ij} \in (0, 1] represents their affinity weight computed via a Gaussian kernel:

    αij=exp⁡(−∥xi−xj∥222σ2)\alpha_{ij} = \exp\left(-\frac{\|x_i - x_j\|_2^2}{2\sigma^2}\right)

    with σ\sigma being the kernel bandwidth parameter.

  3. Knowl 3 — Auxiliary Classifier Generative Adversarial Network for Hyperspectral Classification

    model/method

    In generative adversarial network (GAN)-based hyperspectral image classification, an auxiliary classifier GAN (AC-GAN) configuration is employed to handle multi-class land-cover assignment. The generator GG takes random noise z∼p(z)z \sim p(z) conditioned on a class label cc to generate synthetic hyperspectral samples xfake=G(z,c)x_{\text{fake}} = G(z, c). The discriminator DD acts simultaneously as a source discriminator (real versus synthetic data) and a multi-class softmax classifier.

    The objective function consists of two terms: the log-likelihood of the data source LsL_s and the log-likelihood of the correct class label LcL_c:

    Ls=Exreal[log⁡P(s=real∣xreal)]+Exfake[log⁡P(s=fake∣xfake)]L_s = \mathbb{E}_{x_{\text{real}}}\left[\log P(s = \text{real} \mid x_{\text{real}})\right] + \mathbb{E}_{x_{\text{fake}}}\left[\log P(s = \text{fake} \mid x_{\text{fake}})\right]

    Lc=Exreal,creal[log⁡P(c=creal∣xreal)]+Exfake,cfake[log⁡P(c=cfake∣xfake)]L_c = \mathbb{E}_{x_{\text{real}}, c_{\text{real}}}\left[\log P(c = c_{\text{real}} \mid x_{\text{real}})\right] + \mathbb{E}_{x_{\text{fake}}, c_{\text{fake}}}\left[\log P(c = c_{\text{fake}} \mid x_{\text{fake}})\right]

    During adversarial training, DD is optimized to maximize Ls+LcL_s + L_c, whereas GG is optimized to maximize Lc−LsL_c - L_s.

  4. Knowl 4 — Residual Block Optimization for Deep Hyperspectral Networks

    equation

    In deep networks for hyperspectral classification, residual learning avoids vanishing gradients when increasing network depth by learning a residual mapping G(X)G(X) relative to an identity shortcut.

    Let XX denote the input feature tensor and F(X)F(X) the desired underlying mapping. The mapping is parameterized as:

    F(X)=G(X)+XF(X) = G(X) + X

    For a residual unit comprising two convolutional operations with ReLU activation functions σ(x)=max⁡(0,x)\sigma(x) = \max(0, x):

    G(X)=σ(σ(X∗W1+b1)∗W2+b2)G(X) = \sigma\left(\sigma\left(X * W_1 + b_1\right) * W_2 + b_2\right)

    where W1W_1 and W2W_2 are convolutional filter weights, b1b_1 and b2b_2 are trainable bias vectors, and ∗* denotes the convolution operator.

  5. Knowl 5 — Classification Accuracies of Traditional and Deep Learning Models on the Houston Dataset

    data/table

    The table compares four traditional machine learning methods (Support Vector Machine [SVM], Extended Morphological Profiles [EMP], Joint Sparse Representation [JSR], Edge-Preserving Filtering [EPF]) against six deep learning methods (3-D CNN, CNN with pixel-pair features [CNN-PPF], Gabor-CNN, Siamese CNN [S-CNN], 3-D GAN, and Deep Feature Fusion Network [DFFN]) on the 2013 GRSS Houston dataset (144 bands, 15 classes, 2832 training samples, 12179 test samples). Evaluation metrics are Overall Accuracy (OA, %), Average Accuracy (AA, %), and the Kappa coefficient, averaged over ten independent runs.

    Metric SVM EMP JSR EPF 3D-CNN CNN-PPF Gabor-CNN S-CNN 3D-GAN DFFN
    OA (%) 76.88 79.58 77.12 79.98 82.23 81.38 83.42 82.34 78.16 84.56
    AA (%) 80.29 80.78 78.86 82.21 84.08 84.44 84.17 84.43 79.98 86.18
    Kappa 0.7513 0.7815 0.7528 0.7824 0.8013 0.7991 0.8114 0.8052 0.7616 0.8328

    DFFN achieves the highest OA, AA, and Kappa by combining residual learning with multi-layer feature fusion, while spatial-spectral deep models outperform pixel-wise and shallow spectral-spatial classifiers.

  6. Knowl 6 — Classification Accuracies of Traditional and Deep Learning Models on the University of Pavia Dataset

    data/table

    The table compares four traditional methods (SVM, EMP, JSR, EPF) against six deep learning methods (3D-CNN, CNN-PPF, Gabor-CNN, S-CNN, 3D-GAN, DFFN) on the University of Pavia dataset (103 spectral bands, 9 classes, 200 training samples per class totaling 1800 samples, 40976 test samples). Values reflect the mean performance across ten runs.

    Metric SVM EMP JSR EPF 3D-CNN CNN-PPF Gabor-CNN S-CNN 3D-GAN DFFN
    OA (%) 90.78 96.16 94.43 97.58 96.37 97.63 97.33 97.93 97.81 98.57
    AA (%) 92.14 95.08 92.57 97.72 94.82 97.04 96.62 97.88 96.65 99.16
    Kappa 0.8813 0.9504 0.9349 0.9688 0.9502 0.9690 0.9662 0.9743 0.9697 0.9808

    Deep learning architectures incorporating spatial-spectral contextual information (notably DFFN and S-CNN) attain superior classification accuracies over shallow pixel-wise classifiers and hand-crafted feature extractors.

  7. Knowl 7 — Classification Accuracies of Traditional and Deep Learning Models on the Salinas Dataset

    data/table

    The table compares traditional classifiers (SVM, EMP, JSR, EPF) and deep networks (3D-CNN, CNN-PPF, Gabor-CNN, S-CNN, 3D-GAN, DFFN) on the AVIRIS Salinas dataset (204 spectral bands, 16 classes, 200 training samples per class totaling 3200 samples, 50929 test samples). Reported metrics are averaged over ten independent runs.

    Metric SVM EMP JSR EPF 3D-CNN CNN-PPF Gabor-CNN S-CNN 3D-GAN DFFN
    OA (%) 90.92 96.89 95.33 97.32 97.28 94.87 97.63 97.62 98.22 99.71
    AA (%) 96.18 97.95 94.78 98.88 97.48 98.07 98.04 96.94 97.75 99.78
    Kappa 0.8985 0.9707 0.9307 0.9700 0.9695 0.9404 0.9734 0.9510 0.9793 0.9967

    DFFN attains near-perfect performance (OA of 99.71%, AA of 99.78%, and Kappa of 0.9967), outperforming traditional methods and non-residual deep models.

  8. Knowl 8 — Empirical Comparison of Sample-Efficiency Strategies Under Small Training Regimes

    empirical result

    On the Salinas hyperspectral dataset evaluated across varying numbers of labeled training samples per class (5,10,15,20,25,305, 10, 15, 20, 25, 30), four configurations of a 7-layer convolutional neural network architecture were evaluated:

    1. CNN-Original: Baseline 7-layer CNN with batch normalization, global pooling, and two fully connected layers.
    2. CNN-DA (Data Augmentation): CNN trained with mixture-based virtual sample generation via Gaussian affinity weighting.
    3. CNN-TL (Transfer Learning): CNN pretrained on the Indian Pines dataset (acquired by the same AVIRIS sensor) and fine-tuned on Salinas.
    4. CNN-RL (Residual Learning): CNN augmented with identity shortcut residual blocks.

    Key empirical findings:

    • All three strategies (CNN-DA, CNN-TL, CNN-RL) consistently yield higher Overall Accuracy (OA) than CNN-Original across all sample size regimes (55 to 3030 samples/class).
    • At the lowest sample regime (5 samples per class), transfer learning (CNN-TL) improves OA over CNN-Original by approximately 1%.
    • Across most sample size regimes, residual learning (CNN-RL) achieves the highest overall classification accuracy among all evaluated strategies, demonstrating the effectiveness of identity skip connections in stabilizing feature representations under limited supervision.

Coverage note — No substantial contributed material was omitted.

References

  1. 1.L. Liang, L. Di, L. Zhang, M. Deng, Z. Qin, S. Zhao, and H. Lin, “Estimation of crop lai using hyperspectral vegetation indices and a hybrid inversion method,” Remote Sens. Environ., vol. 165, pp. 123–134, Aug. 2015.
  2. 2.X. Yang and Y. Yu, “Estimating soil salinity under various moisture conditions: An experimental study,” IEEE Trans. Geosci. Remote Sens., vol. 55, no. 5, pp. 2525–2533, May 2017.
  3. 3.N. Yokoya, J. C.-W. Chan, and K. Segl, “Potential of resolution-enhanced hyperspectral data for mineral mapping using simulated enmap and sentinel-2 images,” Remote Sens., vol. 8, no. 3, pp. 172–189, Feb. 2016.
  4. 4.S. Li, R. Dian, L. Fang, and J. M. Bioucas-Dias, “Fusing hyperspectral and multispectral images via coupled sparse tensor factorization,” IEEE Trans. Image Process., vol. 27, no. 8, pp. 4118–4130, Aug. 2018.
  5. 5.S. Zhang, J. Li, Z. Wu, and A. Plaza, “Spatial discontinuity-weighted sparse unmixing of hyperspectral images,” IEEE Trans. Geosci. Remote Sens., vol. 56, no. 10, pp. 5767–5779, Oct. 2018.
  6. 6.P. Ghamisi, N. Yokoya, J. Li, W. Liao, S. Liu, J. Plaza, B. Rasti, and A. Plaza, “Advances in hyperspectral image and signal processing: A comprehensive overview of the state of the art,” IEEE Geosci. Remote Sens. Mag., vol. 5, no. 4, pp. 37–78, Dec. 2017.
  7. 7.Y. Zhong and L. Zhang, “An adaptive artificial immune network for supervised classification of multi-/hyperspectral remote sensing imagery,” IEEE Trans. Geosci. Remote Sens., vol. 50, no. 3, pp. 894–909, Mar. 2012.
  8. 8.F. Melgani and L. Bruzzone, “Classification of hyperspectral remote sensing images with support vector machines,” IEEE Trans. Geosci. Remote Sens., vol. 42, no. 8, pp. 1778–1790, Aug. 2004.
  9. 9.J. Li, J. M. Bioucas-Dias, and A. Plaza, “Semisupervised hyperspectral image segmentation using multinomial logistic regression with active learning,” IEEE Trans. Geosci. Remote Sens., vol. 48, no. 11, pp. 4085–4098, Nov. 2010.
  10. 10.J. Li, J. M. Bioucas-Dias, and A. Plaza, “Spectral–spatial hyperspectral image segmentation using subspace multinomial logistic regression and markov random fields,” IEEE Trans. Geosci. Remote Sens., vol. 50, no. 3, pp. 809–823, Mar. 2012.
  11. 11.B. Du and L. Zhang, “Random-selection-based anomaly detector for hyperspectral imagery,” IEEE Trans. Geosci. Remote Sens., vol. 49, no. 5, pp. 1578–1589, May 2011.
  12. 12.B. Du and L. Zhang, “Target detection based on a dynamic subspace,” Pattern Recog., vol. 47, no. 1, pp. 344–358, Jan. 2014.
  13. 13.G. Licciardi, P. R. Marpu, J. Chanussot, and J. A. Benediktsson, “Linear versus nonlinear PCA for the classification of hyperspectral data based on the extended morphological profiles,” IEEE Geosci. Remote Sens. Lett., vol. 9, no. 3, pp. 447–451, May 2012.
  14. 14.S. Prasad and L. M. Bruce, “Limitations of principal components analysis for hyperspectral target recognition,” IEEE Geosci. Remote Sens. Lett., vol. 5, no. 4, pp. 625–629, Oct. 2008.
  15. 15.A. Villa, J. A. Benediktsson, J. Chanussot, and C. Jutten, “Hyperspectral image classification with independent component discriminant analysis,” IEEE Trans. Geosci. Remote Sens., vol. 49, no. 12, pp. 4865–4876, Dec. 2011.
  16. 16.T. V. Bandos, L. Bruzzone, and G. Camps-Valls, “Classification of hyperspectral images with regularized linear discriminant analysis,” IEEE Trans. Geosci. Remote Sens., vol. 47, no. 3, pp. 862–873, Mar. 2009.
  17. 17.P. Ghamisi et al., “New frontiers in spectral-spatial hyperspectral image classification: The latest advances based on mathematical morphology, markov random fields, segmentation, sparse representation, and deep learning,” IEEE Geosci. Remote Sens. Mag., vol. 6, no. 3, pp. 10–43, Sep. 2018.
  18. 18.L. He, J. Li, C. Liu, and S. Li, “Recent advances on spectral-spatial hyperspectral image classification: An overview and new guidelines,” IEEE Trans. Geosci. Remote Sens., vol. 56, no. 3, pp. 1579–1597, Mar. 2018.
  19. 19.J. A. Benediktsson, J. Palmason, and J. R. Sveinsson, “Classification of hyperspectral data from urban areas based on extended morphological profiles,” IEEE Trans. Geosci. Remote Sens., vol. 43, no. 3, pp. 480–491, Mar. 2005.
  20. 20.J. Li, P. R. Marpu, A. Plaza, J. M. Bioucas-Dias, and J. A. Benediktsson, “Generalized composite kernel framework for hyperspectral image classification,” IEEE Trans. Geosci. Remote Sens., vol. 51, no. 9, pp. 4816–4829, Sep. 2013.
  21. 21.G. Camps-Valls, L. Gomez-Chova, J. Munoz-Maŕı, J. Vila-Frances, and J. Calpe-Maravilla, “Composite kernels for hyperspectral image classification,” IEEE Trans. Geosci. Remote Lett., vol. 3, no. 1, pp. 93–97, Jan. 2006.
  22. 22.Ma. Fauvel, J. Chanussot, and J. A. Benediktsson, “A spatial–spectral kernel-based approach for the classification of remote-sensing images,” Pattern Recog., vol. 45, no. 1, pp. 381–392, Jan. 2012.
  23. 23.L. Fang, S. Li, W. Duan, J. Ren, and J. A. Benediktsson, “Classification of hyperspectral images by exploiting spectral–spatial information of superpixel via multiple kernels,” IEEE Trans. Geosci. Remote Sens., vol. 53, no. 12, pp. 6663–6674, Dec. 2015.
  24. 24.X. Kang, S. Li, and J. A. Benediktsson, “Spectral-spatial hyperspectral image classification with edge-preserving filtering,” IEEE Trans. Geosci. Remote Sens., vol. 52, no. 5, pp. 2666–2677, May 2014.
  25. 25.Y. Chen, N. M. Nasrabadi, and T. D. Tran, “Hyperspectral image classification using dictionary-based sparse representation,” IEEE Trans. Geosci. Remote Sens., vol. 49, no. 10, pp. 3973–3985, Oct. 2011.
  26. 26.L. Fang, S. Li, X. Kang, and J. A. Benediktsson, “Spectral–spatial hyperspectral image classification via multiscale adaptive sparse representation,” IEEE Trans. Geosci. Remote Sens., vol. 52, no. 12, pp. 7738–7749, Dec. 2014.
  27. 27.L. Fang, C. Wang, S. Li, and J. A. Benediktsson, “Hyperspectral image classification via multiple-feature-based adaptive sparse representation,” IEEE Trans. Instrum. Meas., vol. 66, no. 7, pp. 1646–1657, Jul. 2017.
  28. 28.S. Li, T. Lu, L. Fang, X. Jia, and J. A. Benediktsson, “Probabilistic fusion of pixel-level and superpixel-level hyperspectral image classification,” IEEE Trans. Geosci. Remote Sens., vol. 54, no. 12, pp. 7416–7430, Dec. 2016.
  29. 29.T. Lu, S. Li, L. Fang, X. Jia, and J. A. Benediktsson, “From subpixel to superpixel: A novel fusion framework for hyperspectral image classification,” IEEE Trans. Geosci. Remote Sens., vol. 55, no. 8, pp. 4398–4411, Aug. 2017.
  30. 30.L. Fang, N. He, S. Li, P. Ghamisi, and J. A. Benediktsson, “Extinction profiles fusion for hyperspectral images classification,” IEEE Trans. Geosci. Remote Sens., vol. 56, no. 3, pp. 1803–1815, Mar. 2018.
  31. 31.A. Krizhevsky, I. Sutskever, and G. E. Hinton, “Imagenet classification with deep convolutional neural networks,” in Proc. Adv. Neural Inf. Process. Syst., 2012, pp. 1097–1105.
  32. 32.C. Szegedy, W. Liu, Y. Jia, P. Sermanet, S. R., D. Anguelov, D. Erhan, V. Vanhoucke, and A. Rabinovich, “Going deeper with convolutions,” in Proc. IEEE Int. Conf. Comput. Vis. Pattern Recog., 2015, pp. 1–9.
  33. 33.R. Girshick, J. Donahue, T. Darrell, and J. Malik, “Rich feature hierarchies for accurate object detection and semantic segmentation,” in Proc. IEEE Int. Conf. Comput. Vis. Pattern Recog., 2014, pp. 580–587.
  34. 34.A. Bordes, X. Glorot, J. Weston, and Y. Bengio, “Joint learning of words and meaning representations for open-text semantic parsing,” in Proc. Int. Conf. Art. Intell. Stat, 2012, pp. 127–135.
  35. 35.Y. Chen, Z. Lin, X. Zhao, G. Wang, and Y. Gu, “Deep learning-based classification of hyperspectral data,” IEEE J. Sel. Topics Appl. Earth Observ. Remote Sens., vol. 7, no. 6, pp. 2094–2107, Jun. 2014.
  36. 36.P. Ghamisi, J. Plaza, Y. Chen, J. Li, and A. J. Plaza, “Advanced spectral classifiers for hyperspectral images: A review,” IEEE Trans. Geosci. Remote Sens. Maga., vol. 5, no. 1, pp. 8–32, Mar. 2017.
  37. 37.Y. Chen, X. Zhao, and X. Jia, “Spectral-spatial classification of hyperspectral data based on deep belief network,” IEEE J. Sel. Topics Appl. Earth Observ. Remote Sens., vol. 8, no. 6, pp. 2381–2392, Jun. 2015.
  38. 38.Y. Bengio, P. Simard, and P. Frasconi, “Learning long-term dependencies with gradient descent is difficult,” IEEE Trans. Neural Netw., vol. 5, no. 2, pp. 157–166, Mar. 1994.
  39. 39.R. J. Williams and D. Zipser, “A learning algorithm for continually running fully recurrent neural networks,” Neural Comput., vol. 1, no. 2, pp. 270–280, 1989.
  40. 40.P. Rodriguez, J. Wiles, and J. L. Elman, “A recurrent neural network that learns to count,” Connection Sci., vol. 11, no. 1, pp. 5–40, 1999.
  41. 41.S. Hochreiter and J. Schmidhuber, “Long short-term memory,” Neural Comput., vol. 9, no. 8, pp. 1735–1780, 1997.
  42. 42.A. Graves, “Generating sequences with recurrent neural networks,” arXiv preprint, arXiv:1308.0850, 2013.
  43. 43.J. Chung, C Gulcehre, K. Cho, and Y. Bengio, “Empirical evaluation of gated recurrent neural networks on sequence modeling,” arXiv preprint arXiv:1412.3555, 2014.
  44. 44.L. Mou, P. Ghamisi, and X. Zhu, “Deep recurrent neural networks for hyperspectral image classification,” IEEE Trans. Geosci. Remote Sens., vol. 55, no. 7, pp. 3639–3655, Jul. 2017.
  45. 45.I. J. Goodfellow, J. Pouget-Abadie, M. Mirza, B. Xu, D. Warde-Farley, S. Ozair, A. Courville, and Y. Bengio, “Generative adversarial nets,” in Proc. Adv. Neural Inf. Process. Syst., 2014, pp. 2672–2680.
  46. 46.L. Zhu, Y. Chen, P. Ghamisi, and J. A. Benediktsson, “Generative adversarial networks for hyperspectral image classification,” IEEE Trans. Geosci. Remote Sens., vol. 56, no. 9, pp. 5046–5063, Sep. 2018.
  47. 47.A. Radford, L. Metz, and S. Chintala, “Unsupervised representation learning with deep convolutional generative adversarial networks,” arXiv preprint arXiv: 1511.06434v2, 2016.
  48. 48.A. Odena, C. Olah, and J. Shlens, “Conditional image synthesis with auxiliary classifier gans,” arXiv preprint arXiv: 1610.09585, 2016.
  49. 49.Y. Lecun, Y. Bengio, and G. Hinton, “Deep learning,” Nature, vol. 521, no. 7553, pp. 436, 2015.
  50. 50.L. Zhang, L. Zhang, and B. Du, “Deep learning for remote sensing data: A technical tutorial on the state of the art,” IEEE Trans. Geosci. Remote Sens. Mag., vol. 4, no. 2, pp. 22–40, Jun. 2016.
  51. 51.X. Zhu, D. Tuia, L. Mou, G. Song Xia, L. Zhang, F. Xu, and F. Fraundorfer, “Deep learning in remote sensing: A comprehensive review and list of resources,” IEEE Trans. Geosci. Remote Sens. Mag., vol. 5, no. 4, pp. 8–36, Dec. 2017.
  52. 52.Y. Li, H. Zhang, X. Xue, Y. Jiang, and Q. Shen, “Deep learning for remote sensing image classification: A survey,” WIREs Data Mining Knowl. Discov., , no. 12, pp. 1264–1280, May 2018.
  53. 53.P. Liu, H. Zhang, and K. B. Eom, “Active deep learning for classification of hyperspectral images,” IEEE J. Sel. Topics Appl. Earth Observ. Remote Sens., vol. 10, no. 2, pp. 712–724, Feb. 2017.
  54. 54.P. Zhong, Zh. Gong, S. Li, and C. B. Schonlieb, “Learning to diversify deep belief networks for hyperspectral image classification,” IEEE Trans. Geosci. Remote Sens., vol. 55, no. 6, pp. 3516–3530, Jun. 2017.
  55. 55.W. Hu, Y. Huang, L. Wei, F. Zhang, and H. Li, “Deep convolutional neural networks for hyperspectral image classification,” J. Sensors, vol. 2015, Jan. 2015, Art. no. 258619.
  56. 56.Y. Chen, H. Jiang, C. Li, X. Jia, and P. Ghamisi, “Deep feature extraction and classification of hyperspectral images based on convolutional neural networks,” IEEE Trans. Geosci. Remote Sens., vol. 54, no. 10, pp. 6232–6251, Oct. 2016.
  57. 57.J. M. Haut, M. E. Paoletti, J. Plaza, J. Li, and A. Plaza, “Active learning with convolutional neural networks for hyperspectral image classification using a new bayesian approach,” IEEE Trans. Geosci. Remote Sens., vol. 56, no. 11, pp. 6440–6461, Nov. 2018.
  58. 58.X. Yang, Y. Ye, X. Li, R. Y. K. Lau, X. Zhang, and X. Huang, “Hyperspectral image classification with deep learning models,” IEEE Trans. Geosci. Remote Sens., vol. 56, no. 9, pp. 5408–5423, Sep. 2018.
  59. 59.Y. Zhan, D. Hu, Y. Wang, and X. Yu, “Semisupervised hyperspectral image classification based on generative adversarial networks,” IEEE Geosci. Remote Lett., vol. 15, no. 2, pp. 212–216, Feb. 2018.
  60. 60.H. Wu and S. Prasad, “Convolutional recurrent neural networks forhyperspectral data classification,” Remote Sens., vol. 9, no. 3, pp. 298–317, Mar. 2017.
  61. 61.W. Li, G. Wu, F. Zhang, and Q. Du, “Hyperspectral image classification using deep pixel-pair features,” IEEE Trans. Geosci. Remote Sens., vol. 55, no. 2, pp. 844–853, Feb. 2017.
  62. 62.S. Tariyal, H. K. Aggarwal, and A. Majumdar, “Greedy deep dictionary learning for hyperspectral image classification,” in Proc. IEEE GRSS Workshop Hyperspectral Image Signal Process.: Evol. Remote Sens., 2016, pp. 21–24.
  63. 63.V. Singhal, H. K. Aggarwal, S. Tariyal, and A. Majumdar, “Discriminative robust deep dictionary learning for hyperspectral image classification,” IEEE Trans. Geosci. Remote Sens., vol. 55, no. 9, pp. 5274–5283, Sep. 2017.
  64. 64.J. A. Benediktsson and P. Ghamisi, Spectral-spatial classification of hyperspectral remote sensing images, Artech House, INC, Boston, USA, 2015.
  65. 65.J. Yue, W. Zhao, S. Mao, and H. Liu, “Spectral-spatial classification of hyperspectral images using deep convolutional neural networks,” Remote Sens. Lett., vol. 6, no. 6, pp. 468–477, May 2015.
  66. 66.K. Makantasis, K. Karantzalos, A. Doulamis, and N. Doulamis, “Deep supervised learning for hyperspectral data classification through convolutional neural networks,” in Proc. IEEE Geosci. Remote Sens. Symp., 2015, pp. 4959–4962.
  67. 67.L. Fang, Z. Liu, and W. Song, “Deep hashing neural networks for hyperspectral image feature extraction,” IEEE Geosci. Remote Sens. Lett., DOI: 10.1109/LGRS.2019.2899823, to be published, 2019.
  68. 68.H. Liang and Q. Li, “Hyperspectral imagery classification using sparse representations of convolutional neural network features,” Remote Sens., vol. 8, no. 2, pp. 99–114, Jan. 2016.
  69. 69.G. Cheng, Z. Li, J. Han, X. Yao, and L. Guo, “Exploring hierarchical convolutional features for hyperspectral image classification,” IEEE Trans. Geosci. Remote Sens., vol. 56, no. 11, pp. 6712–6722, Nov. 2018.
  70. 70.W. Zhao and S Du, “Spectral-spatial feature extraction for hyperspectral image classification: A dimension reduction and deep learning approach,” IEEE Trans. Geosci. Remote Sens., vol. 54, no. 8, pp. 4544–4554, Aug. 2016.
  71. 71.L. Jiao, M. Liang, H. Chen, S. Yang, H. Liu, and X. Cao, “Deep fully convolutional network-based spatial distribution prediction for hyperspectral image classification,” IEEE Trans. Geosci. Remote Sens., vol. 55, no. 10, pp. 5585–5599, Oct. 2017.
  72. 72.J. Long, E. Shelhamer, and T. Darrell, “Fully convolutional networks for semantic segmentation,” in Proc. IEEE Int. Conf. Comput. Vis. Pattern Recog., 2015, pp. 3431–3440.
  73. 73.X. Ma, H. Wang, and J. Geng, “Spectral-spatial classification of hyperspectral image based on deep auto-encoder,” IEEE J. Sel. Topics Appl. Earth Observ. Remote Sens., vol. 9, no. 9, pp. 4073–4085, Sep. 2016.
  74. 74.X. Zhou, S. Li, F. Tang, K. Qin, S. Hu, and S. Liu, “Deep learning with grouped features for spatial spectral classification of hyperspectral images,” IEEE Geosci. Remote Sens. Lett., vol. 14, no. 1, pp. 97–101, Jan. 2017.
  75. 75.X. Sun, F. Zhou, J. Dong, F. Gao, Q. Mu, and X. Wang, “Encoding spectral and spatial context information for hyperspectral image classification,” IEEE Geosci. Remote Sens. Lett., vol. 14, no. 12, pp. 2250–2254, Dec. 2017.
  76. 76.Q. Lv, X. Niu, Y. Dou, J. Xu, and Y. Lei, “Classification of hyperspectral remote sensing image using hierarchical local-receptive-field-based extreme learning machine,” IEEE Geosci. Remote Sens. Lett., vol. 13, no. 3, pp. 434–438, Mar. 2016.
  77. 77.S. Mei, J. Ji, J. Hou, X. Li, and Q. Du, “Learning sensor-specific spatial-spectral features of hyperspectral images via convolutional neural networks,” IEEE Trans. Geosci. Remote Sens., vol. 55, no. 8, pp. 4520–4533, Aug. 2017.
  78. 78.X. Kang, C. Li, S. Li, and H. Lin, “Classification of hyperspectral images by gabor filtering based deep network,” IEEE J. Sel. Topics Appl. Earth Observ. Remote Sens., vol. 11, no. 4, pp. 1166–1178, Apr. 2018.
  79. 79.Y. Chen, L. Zhu, P. Ghamisi, X. Jia, G. Li, and L. Tang, “Hyperspectral images classification with gabor filtering and convolutional neural network,” IEEE Geosci. Remote Sens. Lett., vol. 14, no. 12, pp. 2355–2359, Dec. 2017.
  80. 80.E. Aptoula, M. Can Ozdemir, and B. Yanikoglu, “Deep learning with attribute profiles for hyperspectral image classification,” IEEE Geosci. Remote Sens. Lett., vol. 13, no. 12, pp. 1970–1974, Dec. 2016.
  81. 81.P. Ghamisi, B. Hofle, and X. Zhu, “Hyperspectral and lidar data fusion using extinction profiles and deep convolutional neural network,” IEEE J. Sel. Topics Appl. Earth Observ. Remote Sens., vol. 10, no. 6, pp. 3011–3024, Jun. 2017.
  82. 82.B. Pan, Z. Shi, and X. Xu, “R-VCANet: A new deep-learning-based hyperspectral image classification method,” IEEE J. Sel. Topics Appl. Earth Observ. Remote Sens., vol. 10, no. 5, pp. 1975–1986, May 2017.
  83. 83.N. He, M. E. Paoletti, Haut J. M., L. Fang, S. Li, A. Plaza, and J. Plaza, “Feature extraction with multiscale covariance maps for hyperspectral image classification,” IEEE Trans. Geosci. Remote Sens., vol. 57, no. 2, pp. 755–769, Feb. 2019.
  84. 84.Y. Chen, C. Li, P. Ghamisi, X. Jia, and Y. Gu, “Deep fusion of remote sensing data for accurate classification,” IEEE Geosci. Remote Sens. Lett., vol. 14, no. 8, pp. 1253–1257, Aug. 2017.
  85. 85.M. Zhang, W. Li, and Q. Du, “Diverse region-based CNN for hyperspectral image classification.,” IEEE Trans. Image Process., vol. 27, no. 6, pp. 2623, Jun. 2018.
  86. 86.X. Cao, F. Zhou, L. Xu, D. Meng, Z. Xu, and J Paisley, “Hyperspectral image classification with markov random fields and a convolutional neural network,” IEEE Trans. Image Process., vol. 27, no. 5, pp. 2354–2367, May 2018.
  87. 87.W. Song, S. Li, L. Fang, and T. Lu, “Hyperspectral image classification with deep feature fusion network,” IEEE Trans. Geosci. Remote Sens., vol. 56, no. 6, pp. 3173–3184, Jun. 2018.
  88. 88.W. Song, S. Li, and Y. Li, “Hyperspectral images classification with hybrid deep residual network,” in Proc. IEEE Geosci. Remote Sens. Symp., 2017, pp. 2235–2238.
  89. 89.B. Liu, X. Yu, P. Zhang, A. Yu, Q. Fu, and X. Wei, “Supervised deep feature extraction for hyperspectral image classification,” IEEE Trans. Geosci. Remote Sens., vol. 56, no. 4, pp. 1909–1921, Apr. 2018.
  90. 90.J. Zhu, L. Fang, and P. Ghamisi, “Deformable convolutional neural networks for hyperspectral image classification,” IEEE Geosci. Remote Sens. Lett., vol. 15, no. 8, pp. 1254–1258, Aug. 2018.
  91. 91.L. Fang, G. Liu, S. Li, P. Ghamisi, and J. A. Benediktsson, “Hyperspectral image ilassification with squeeze multibias network,” IEEE Trans. Geosci. Remote Sens., vol. 57, no. 3, pp. 1291–1301, Mar. 2019.
  92. 92.Y. Li, H. Zhang, and Q. Shen, “Spectral-spatial classification of hyperspectral imagery with 3D convolutional neural network,” Remote Sens., vol. 9, no. 1, pp. 67–87, Jan. 2017.
  93. 93.Z. Zhong, J. Li, Z. Luo, and M. Chapman, “Spectral-spatial residual network for hyperspectral image classification: A 3-D deep learning framework,” IEEE Trans. Geosci. Remote Sens., vol. 56, no. 2, pp. 847–856, Feb. 2018.
  94. 94.A. B. Hamida, A. Benoit, P. Lambert, and C. B. Amar, “3-D deep learning approach for remote sensing image classification,” IEEE Trans. Geosci. Remote Sens., vol. 56, no. 8, pp. 4420–4434, Aug. 2018.
  95. 95.M. E. Paoletti, J. M Haut, J Plaza, and A Plaza, “A new deep convolutional neural network for fast hyperspectral image classification,” ISPRS J. of Photogramm. and Remote Sens., vol. 145, pp. 120–147, 2018.
  96. 96.J. Li, X. Zhao, Y. Li, Q. Du, B. Xi, and J. Hu, “Classification of hyperspectral imagery using a new fully convolutional neural network,” IEEE Geosci. Remote Sens. Lett., vol. 15, no. 2, pp. 292–296, Feb. 2018.
  97. 97.L. Mou, P. Ghamisi, and X. Zhu, “Unsupervised spectral-spatial feature learning via deep residual conv-deconv network for hyperspectral image classification,” IEEE Trans. Geosci. Remote Sens., vol. 56, no. 1, pp. 391–406, Jan. 2018.
  98. 98.H. Lee and H. Kwon, “Going deeper with contextual cnn for hyperspectral image classification,” IEEE Trans. Image Process., vol. 26, no. 10, pp. 4843–4855, Oct. 2017.
  99. 99.M. E. Paoletti, J. M. Haut, R. Fernandez-Beltran, J. Plaza, A. J. Plaza, and F. Pla, “Deep pyramidal residual networks for spectral-spatial hyperspectral image classification,” IEEE Trans. Geosci. Remote Sens., vol. 57, no. 2, pp. 740–754, Feb. 2019.
  100. 100.K. He, X. Zhang, S. Ren, and J. Sun, “Deep residual learning for image recognition,” in Proc. IEEE Int. Conf. Comput. Vis. Pattern Recog., 2016, pp. 770–778.
  101. 101.M. Zhang, M. Gong, Y. Mao, J. Li, and Y. Wu, “Unsupervised feature extraction in hyperspectral images based on wasserstein generative adversarial network,” IEEE Trans. Geosci. Remote Sens., DOI: 10.1109/TGRS.2018.2876123, to be published, 2019.
  102. 102.S. Sabour, N. Frosst, and G. E. Hinton, “Dynamic routing between capsules,” in in Proc. Adv. Neural Inf. Process. Syst, 2017, pp. 3856–3866.
  103. 103.M. E. Paoletti, J. M. Haut, R. Fernandez-Beltran, J. Plaza, A. J. Plaza, J. Li, and F. Pla, “Capsule networks for hyperspectral image classification,” IEEE Trans. Geosci. Remote Sens., DOI: 10.1109/TGRS.2018.2871782, to be published,2019.
  104. 104.R. Kemker and C. Kanan, “Self-taught feature learning for hyperspectral image classification,” IEEE Trans. Geosci. Remote Sens., vol. 55, no. 5, pp. 2639–2705, May 2017.
  105. 105.H. Wu and S. Prasad, “Semi-supervised deep learning using pseudo labels for hyperspectral image classification,” IEEE Trans. Image Process., vol. 27, no. 3, pp. 1259–1270, Mar. 2018.
  106. 106.R. Hang, Q. Liu, D. Hong, and P. Ghamisi, “Cascaded recurrent neural networks for hyperspectral image classification,” IEEE Trans. Geosci. Remote Sens., DOI: 10.1109/TGRS.2019.2899129, to be published, 2019.
  107. 107.J. Yang, Yong Q. Zhao, and C.-W. Chan, “Learning and transferring deep joint spectral-spatial features for hyperspectral classification,” IEEE Trans. Geosci. Remote Sens., vol. 55, no. 8, pp. 4729–4742, Aug. 2017.
  108. 108.X. Xu, W. Li, Q. Ran, Q. Du, L. Gao, and B. Zhang, “Multisource remote sensing data classification based on convolutional neural network,” IEEE Trans. Geosci. Remote Sens., vol. 56, no. 2, pp. 937–949, Feb. 2018.
  109. 109.S. Hao, W. Wang, Y. Ye, T. Nie, and L. Bruzzone, “Two-stream deep architecture for hyperspectral image classification,” IEEE Trans. Geosci. Remote Sens., vol. 56, no. 4, pp. 2349–2361, Apr. 2018.
  110. 110.B. Pan, Z. Shi, N. Zhang, and S. Xie, “Hyperspectral image classification based on nonlinear spectral-spatial network,” IEEE Geosci. Remote Sens. Lett., vol. 13, no. 12, pp. 1782–1786, Dec. 2016.
  111. 111.C. Tao, H. Pan, Y. Li, and Z. Zou, “Unsupervised spectral-spatial feature learning with stacked sparse autoencoder for hyperspectral imagery classification,” IEEE Geosci. Remote Sens. Lett., vol. 12, no. 12, pp. 2438–2442, Dec. 2015.
  112. 112.A. Santara, K. Mani, P. Hatwar, A. Singh, A. Garg, K. Padia, and P. Mitra, “BASS Net: Band-adaptive spectral-spatial feature learning neural network for hyperspectral image classification,” IEEE Trans. Geosci. Remote Sens., vol. 55, no. 9, pp. 5293–5301, Sep. 2017.
  113. 113.J. Yosinski, J. Clune, Y. Bengio, and H. Lipson, “How transferable are features in deep neural networks?,” in Proc. Adv. Neural Inf. Process. Syst., 2014, pp. 3320–3328.
  114. 114.F. Hu, G.-S. Xia, J. Hu, and L. Zhang, “Transferring deep convolutional neural networks for the scene classification of high-resolution remote sensing imagery,” Remote Sens., vol. 7, no. 11, pp. 14680–14707, Nov. 2015.
  115. 115.D. Marmanis, M. Datcu, T. Esch, and U. Stilla, “Deep learning earth observation classification using imagenet pretrained networks,” IEEE Geosci. Remote Sens. Lett., vol. 13, no. 1, pp. 105–109, Jan. 2016.
  116. 116.X. Zhang, Y. Liang, C. Li, H. Ning, L. Jiao, and H. Zhou, “Recursive autoencoders-based unsupervised feature learning for hyperspectral image classification,” IEEE Geosci. Remote Sens. Lett., vol. 14, no. 11, pp. 1928–1932, Nov. 2017.
  117. 117.A. Romero, C. Gatta, and G. Camps-Valls, “Unsupervised deep feature extraction for remote sensing image classification,” IEEE Trans. Geosci. Remote Sens., vol. 54, no. 3, pp. 1349–1362, Mar. 2016.
  118. 118.S. Ioffe and C. Szegedy, “Batch normalization: Accelerating deep network training by reducing internal covariate shift,” arXiv preprint arXiv: 1502.03167, pp. 448–456, 2015.
  119. 119.C.-C. Chang and C.-J Lin, “Libsvm: A library for support vector machines,” ACM Trans. Intell. Syst. Technol., vol. 2, no. 3, pp. 27:1–27:27, Jul. 2011.

Citation

MLA
Li, S., et al. “Deep Learning for Hyperspectral Image Classification: An Overview”. IEEE Transactions on Geoscience and Remote Sensing, vol. 57, no. 9, 2019, pp. 6690–709, https://doi.org/10.1109/TGRS.2019.2907932.
APA
Li, S., Song, W., Fang, L., Chen, Y., Ghamisi, P., & Benediktsson, J. A. (2019). Deep Learning for Hyperspectral Image Classification: An Overview. IEEE Transactions on Geoscience and Remote Sensing, 57(9), 6690–6709. https://doi.org/10.1109/TGRS.2019.2907932
Chicago
Li, S., W. Song, L. Fang, Y. Chen, P. Ghamisi, and J. A. Benediktsson. 2019. “Deep Learning for Hyperspectral Image Classification: An Overview”. IEEE Transactions on Geoscience and Remote Sensing 57 (9): 6690–6709. https://doi.org/10.1109/TGRS.2019.2907932.
Harvard
Li, S. et al. (2019) “Deep Learning for Hyperspectral Image Classification: An Overview”, IEEE Transactions on Geoscience and Remote Sensing, 57(9), pp. 6690–6709. Available at: https://doi.org/10.1109/TGRS.2019.2907932.
Vancouver
1. Li S, Song W, Fang L, Chen Y, Ghamisi P, Benediktsson JA (2019) Deep Learning for Hyperspectral Image Classification: An Overview. IEEE Transactions on Geoscience and Remote Sensing 57:6690–6709

BibTeX

@article{Li_2019, title={Deep Learning for Hyperspectral Image Classification: An Overview}, volume={57}, ISSN={1558-0644}, url={http://dx.doi.org/10.1109/TGRS.2019.2907932}, DOI={10.1109/tgrs.2019.2907932}, number={9}, journal={IEEE Transactions on Geoscience and Remote Sensing}, publisher={Institute of Electrical and Electronics Engineers (IEEE)}, author={Li, Shutao and Song, Weiwei and Fang, Leyuan and Chen, Yushi and Ghamisi, Pedram and Benediktsson, Jon Atli}, year={2019}, month=Sept, pages={6690–6709} }
Metadata:Crossref

Access the Paper

This paper is available from its original source. Click below to access the PDF.

Open PDF