Generative Adversarial Networks for Hyperspectral Image Classification

By Lin Zhu, Yushi Chen, Pedram Ghamisi, Jon Atli Benediktsson. IEEE Transactions on Geoscience and Remote Sensing, 2018

584 citations recorded by the source corpus.

A generative adversarial network (GAN) usually contains a generative network and a discriminative network in competition with each other. The GAN has shown its capability in a variety of applications. In this paper, the usefulness and effectiveness of GAN for classification of hyperspectral images (HSIs) are explored for the first time. In the proposed GAN, a convolutional neural network (CNN) is designed to discriminate the inputs and another CNN is used to generate so-called fake inputs. The aforementioned CNNs are trained together: the generative CNN tries to generate fake inputs that are as real as possible, and the discriminative CNN tries to classify the real and fake inputs. This kind of adversarial training improves the generalization capability of the discriminative CNN, which is really important when the training samples are limited. Specifically, we propose two schemes: 1) a well-designed 1D-GAN as a spectral classifier and 2) a robust 3D-GAN as a spectral–spatial classifier. Furthermore, the generated adversarial samples are used with real training samples to fine-tune the discriminative CNN, which improves the final classification performance. The proposed classifiers are carried out on three widely used hyperspectral data sets: Salinas, Indiana Pines, and Kennedy Space Center. The obtained results reveal that the proposed models provide competitive results compared to the state-of-the-art methods. In addition, the proposed GANs open new opportunities in the remote sensing community for the challenging task of HSI classification and also reveal the huge potential of GAN-based methods for the analysis of such complex and inherently nonlinear data.

Subfields: Computer Vision and Pattern Recognition, Media Technology

View Paper: https://www.semanticscholar.org/paper/d4a560b617175d20c9dbd063c5dc2e8f7478e2a6

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

In Media Technology

Citations in the corpus

SpectralFormer: Rethinking Hyperspectral Image Classification With Transformers (2021, methodology): “In [19], GANs were improved, making them applicable to the HS image classification task, with the input of three PCA components and random noises.”
Research Progress on Few-Shot Learning for Remote Sensing Image Interpretation (2021, methodology): “When the GAN was first applied to the hyperspectral feature classification task, it was used as a regularization to solve the overfitting problem under the small-sample condition, and a collaborative training algorithm was proposed to fuse the adversarial samples with the real samples to improve the”
Vision Transformers for Remote Sensing Image Classification (2021, background): “Other approaches based on recurrent neural networks (RNNs) [23], generative adversarial networks (GANs) [24,25], graph convolutional networks (GCNs) [26], and long- short-term memory (LSTM) [27] have been introduced also.”
Spectral–Spatial Morphological Attention Transformer for Hyperspectral Image Classification (2023, background): “Generative adversarial networks (GANs), on the other hand, may help with mitigating the class-imbalance problem in HSI classification [41], [42].”
Spatial-Spectral Transformer for Hyperspectral Image Classification (2021, methodology): “After that, several deep learning models such as the deep belief network [20], CNN [21,22], recurrent neural network [23,24], generative adversarial network [25,26], and capsule network [27,28] were investigated for HSI classification and obtained good classification performance.”
Deep Learning for Multigrade Brain Tumor Classification in Smart Healthcare Systems: A Prospective Survey (2020, background): “Generative adversarial networks (GANs) are among the advanced data-enrichment techniques and are widely used for many applications in diverse domains, such as medical image synthesis [127], [128], compressive sensing MRI [129], [130], superresolution [130], security [131], classification [132], imag”
Multimodal Fusion Transformer for Remote Sensing Image Classification (2022, unknown): “On the other hand, the generative adversarial network also provides a solution for the classification of HSIs containing a class-wise imbalanced number of samples [46], [47].”
Self-Supervised Learning in Remote Sensing: A review (2022, background): “[118] presented a first work to explore the potential of GAN for HSI classification.”
Spectral–Spatial Feature Tokenization Transformer for Hyperspectral Image Classification (2022, background): “generative adversarial networks (GANs) [51], [52], capsule networks [53], and graph convolutional networks (GCNs) [54] have also been introduced for HSI classification.”
Deep Learning for Classification of Hyperspectral Data: A Comparative Review (2019, background): “However, thanks to the theoretical improvements to the GAN framework, deep generative models are now able to synthesize hyperspectral pixels [81] and even small hyperspectral cubes using a 3D CNN [82] from scratch.”
Hyperspectral Image Classification—Traditional to Deep Models: A Survey for Future Prospects (2021, methodology): “Similarly, [264] developed a CNN-based 1D-GAN and 3D-GAN architectures to enhance the classification performance.”
Deep Learning for Land Use and Land Cover Classification Based on Hyperspectral and Multispectral Earth Observation Data: A Review (2020, methodology): “Reference [104] use 1D-CNN in a generative adversarial network (1D-GAN) to generate fake spectral data, and also as a discriminator to classify the spectral features.”
Deep Unsupervised Embedding for Remotely Sensed Images Based on Spatially Augmented Momentum Contrast (2020, methodology): “In more detail, Tile2Vec consistently obtains the second best performance, followed by the pre-trained CNN + PCA, DCGAN, and MARTA GAN.”
Deep Learning for Hyperspectral Image Classification: An Overview (2019, methodology): “Discriminate approaches attempt to model the dependence of labels (i.e., 5 y) on training data (i.e., x), which can be used for predicting y from x. GANs are the relatively new kind of models, which contains a generative model G and a discriminative model D [45].”