Abstract:In recent years, deep learning-based methods have exhibited remarkable performance in the field of hyperspectral image (HSI) classification. However, conventional supervised methods heavily rely on a substantial number of labeled samples. Self-supervised learning, as a prominent unsupervised representation learning technique, offers the potential to extract valuable information from unlabeled data. In this article, we introduce a novel unsupervised approach called self-supervised learning with the multiscale densely connected network (SS-MSDCNet) to make full use of unlabeled samples for HSI classification. First, a two-stream structure was designed to generate more positive pairs, which enables the contrast self-supervised training to learn more useful information from unlabeled data. Subsequently, a data augmentation technique based on spectral splitting was proposed to coordinate the two-stream structure of SS-MSDCNet, enhancing spectral information expression. The backbone of the proposed approach is the multiscale densely connected network (MSDCNet), which obtains input HSIs with various spatial scales by removing peripheral pixels from the original input and subsequently extracts multiscale spatial-spectral features using 3-D densely connected modules and 3-D spatial attention modules. The 3-D densely connected module effectively harnesses multiscale features extracted by various convolutional layers, while the 3-D spatial attention module enhances the network's focus on features conducive to accurate classification. To validate the efficacy of our approach, we conducted extensive experiments using four distinct HSI datasets. The results unequivocally demonstrate that SS-MSDCNet outperforms several well-established supervised and unsupervised classification methods. Furthermore, we designed a transfer experiment to confirm SS-MSDCNet's robust generalization capabilities. The code is available at https://github.com/mrblank99/SS-MSDCNet.

Distance-based Hyperspectral Open-Set Classification of Deep Neural Networks

Hyperspectral Open Set Classification Towards Deep Networks Based on Boxplot

Hyperspectral Open Set Classification with Unknown Classes Rejection Towards Deep Networks

Spatial Anomaly Detection in Hyperspectral Imaging Using Optical Neural Networks

Spectral-Spatial Latent Reconstruction for Open-Set Hyperspectral Image Classification

Orientational Clustering Learning for Open-Set Hyperspectral Image Classification

Neural Stochastic Differential Equation for Hyperspectral Image Classification

A Prototype Network for Hyperspectral Image Open-Set Classification Based on Feature Invariance and Weighted Pearson Distance Measurement

A Semi-Supervised Learning Method for Hyperspectral-Image Open Set Classification

Generative Adversarial Network and Mutual-Point Learning Algorithm for Few-Shot Open-Set Classification of Hyperspectral Images

Spectral-Spatial Evidential Learning Network for Open-Set Hyperspectral Image Classification

Classification Based On Deep Convolutional Neural Networks With Hyperspectral Image

Hyperspectral Image Classification Using Set-To-Set Distance

An End-to-End Hyperspectral Image Classification Method Using Deep Convolutional Neural Network with Spatial Constraint

Wide and Deep Fourier Neural Network for Hyperspectral Remote Sensing Image Classification

Deep Learning for Hyperspectral Image Classification: An Overview

Learning a Deep Similarity Network for Hyperspectral Image Classification

Few‐shot real scenes convolutional anchor clustering twin network for hyperspectral image classification

Hyperspectral Image Classification Based on Spatial and Spectral Kernels Generation Network

Self-Supervised Learning With Multiscale Densely Connected Network for Hyperspectral Image Classification

HSI-CNN: A Novel Convolution Neural Network for Hyperspectral Image