Abstract:Hyperspectral image (HSI) classification is an important topic in remote sensing tasks, aiming to exploit the spectral and spatial information in HSI to assign category labels to each pixel. Recently, numerous deep learning (DL) methods, such as convolutional neural network (CNN) and transformer, have been introduced and achieved promising results. These DL-based models tend to use patch-based input patterns, where a patch is extracted around a central pixel and the neighboring pixels serve as auxiliary roles. However, compared to developing novel models, the characteristics of the patch-based HSI classification framework have not been adequately explored. Most existing patch-based DL methods tend to overlook the potential relationships between the central pixel and its neighboring pixels. In addition, the unique spectral characteristics of HSI (e.g., adjacent bands possess higher correlation) require special attention. Moreover, the design of feature extractors for HSIs appears to be constrained primarily to the spatial and spectral domains. To cope with the above problems, we propose an HSI classification specifically designed model (HSIC-SDM), which encompasses three branches—frequency feature learning, spatial feature learning, and spectral feature learning. The frequency feature learning branch employs a Fourier-based feature extractor to acquire representations in the frequency domain. The spatial feature learning branch emphasizes the significance of central pixel and edge information. The spectral feature learning branch concerns the high correlations in the adjacent spectral bands. Multidomain features derived from three branches provide diverse perspectives for the final classification. Extensive experiments on three datasets demonstrate the superiority of the proposed method over the state-of-the-art (SOTA) compared methods. The code of our work will be available publicly at https://github.com/Tikiten/HSIC-SDM.

Deep-Learning-Based Active Hyperspectral Imaging Classification Method Illuminated by the Supercontinuum Laser

Deep Learning for Hyperspectral Image Classification: An Overview

Superpixel-Based Long-Range Dependent Network for High-Resolution Remote-Sensing Image Classification

Superpixel Guided Deep-Sparse-Representation Learning for Hyperspectral Image Classification

Hyperspectral image classification via active learning and broad learning system

Deep Convolutional Networks with Superpixel Segmentation for Hyperspectral Image Classification

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

Hyperspectral Image Classification Based on Deep Auto-Encoder and Hidden Markov Random Field.

A Hybrid Self-supervised Learning Framework For Hyperspectral Image Classification

Self-Supervised Feature Learning Based on Spectral Masking for Hyperspectral Image Classification

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

Deep Latent Spectral Representation Learning-Based Hyperspectral Band Selection for Target Detection

Hyperspectral Anomaly Detection Using Deep Learning: A Review

Hyperspectral image classification based on superpixel merging and broad learning system

Hyperspectral Image Classification—Traditional to Deep Models: A Survey for Future Prospects

Hyper-voxel Based Deep Learning for Hyperspectral Image Classification

Fortifying Centers and Edges: Multidomain Feature Learning Meets Hyperspectral Image Classification

Deep Convolutional Neural Networks for Hyperspectral Image Classification

Active Deep Densely Connected Convolutional Network for Hyperspectral Image Classification

Active-Learning-Incorporated Deep Transfer Learning for Hyperspectral Image Classification