Abstract:Hyperspectral image (HSI) classification suffers from two serious problems, one is the limited labeled pixels, and the other is the class imbalance problem. As a result, the number of labeled pixels in many categories is not sufficient to characterize the spectral-spatial information, and train a satisfying deep model. By making full use of the information of unlabeled pixels, semi-supervised methods can provide better classification performance in the case of limited labeled pixels. However, they do not take into account the imbalance in the HSI data. As a method of data enhancement, generative adversarial networks focus on the above two problems and have also been widely used for the task of the HSI classification. In this work, we propose a distance constraints-based generative adversarial networks (DGAN) method for HSI classification to address these two problems. The DGAN employs the convolution autoencoder (AE) to extract the latent features of the HSI samples, and considers the reconstructed samples from the AE as the real samples for the later classifier and discriminator. In addition, the DGAN uses two distance constraints to solve the problems of the few labeled samples and class imbalance, the one latent-data distance constraint enforcing the generator to generate HSI samples for each class (especially the minority class), another discriminator-score distance constraint guiding the generator to synthesize samples that resemble the real HSI samples. Finally, the generated samples are combined classwise with the reconstructed samples and the real HSI samples to learn the parameters of the classifier and discriminator. Experimental results show that our method achieves state-of-the-art performance in terms of overall accuracy (OA) when trained with only 0.5%-4% of data sets from Indian Pines, Pavia University, and Botswana. Specifically, our method demonstrates improvements of 5.48%, 8.79%, and 0.91% on these three datasets, respectively. It reveals the gre- t potential of the DGAN model in generating the HSI samples for each class, which contributes to improving the classification performance of the HSI data.

Hyperspectral Target Detection Based on One-Dimensional Generative Adversarial Network

Hyperspectral Target Detection with an Auxiliary Generative Adversarial Network.

Compressive Sensing Ghost Imaging Object Detection Using Generative Adversarial Networks

Robust Signature-Based Hyperspectral Target Detection Using Dual Networks

A Representation-Learning-Based Graph and Generative Network for Hyperspectral Small Target Detection

Hyperspectral Image Classification Based on Generative Adversarial Networks with Feature Fusing and Dynamic Neighborhood Voting Mechanism

Hybrid Spatial-Spectral Generative Adversarial Network for Hyperspectral Image Classification

Manifold Regularized Sparsity Model for Hyperspectral Target Detection

Generative Adversarial Networks and Probabilistic Graph Models for Hyperspectral Image Classification

3-D Auxiliary Classifier GAN for Hyperspectral Anomaly Detection via Weakly Supervised Learning

One-Step Detection Paradigm for Hyperspectral Anomaly Detection via Spectral Deviation Relationship Learning

Distance Constraint-Based Generative Adversarial Networks for Hyperspectral Image Classification

Semisupervised hyperspectral image classification based on generative adversarial networks and spectral angle distance

Single-Pixel Infrared Hyperspectral Imaging via Physics-Guided Generative Adversarial Networks

One-Stream Neural Network Based on Multi-Level Features Fusion for Hyperspectral Image Change Detection

GAN-STD: small target detection based on generative adversarial network

Hyperspectral Target Detection via Spectral Aggregation and Separation Network With Target Band Random Mask

Generating images for Supervised Hyperspectral Image Classification with Generative Adversarial Nets

Generative Adversarial Capsule Network With ConvLSTM for Hyperspectral Image Classification

An Optimized Training Method for GAN-Based Hyperspectral Image Classification

Target detection algorithm based on super- resolution color remote sensing image reconstruction