We can't find the internet
Attempting to reconnect
Something went wrong!
Hang in there while we get back on track
3D Fourier-based Global Feature Extraction for Hyperspectral Image Classification
Summary
Researchers proposed HGFNet, a hybrid deep learning architecture combining 3D convolutional local feature extraction with three complementary Fourier transform strategies for hyperspectral image classification, achieving efficient long-range spectral-spatial modeling and improved discrimination of underrepresented classes via an adaptive focal loss mechanism.
Hyperspectral image classification (HSIC) has been significantly advanced by deep learning methods that exploit rich spatial-spectral correlations. However, existing approaches still face fundamental limitations: transformer-based models suffer from poor scalability due to the quadratic complexity of self-attention, while recent Fourier transform-based methods typically rely on 2D spatial FFTs and largely ignore critical inter-band spectral dependencies inherent to hyperspectral data. To address these challenges, we propose Hybrid GFNet (HGFNet), a novel architecture that integrates localized 3D convolutional feature extraction with frequency-domain global filtering via GFNet-style blocks for efficient and robust spatial-spectral representation learning. HGFNet introduces three complementary frequency transforms tailored to hyperspectral imagery: Spectral Fourier Transform (a 1D FFT along the spectral axis), Spatial Fourier Transform (a 2D FFT over spatial dimensions), and Spatial-Spatial Fourier Transform (a 3D FFT jointly over spectral and spatial dimensions), enabling comprehensive and high-dimensional frequency modeling. The 3D convolutional layers capture fine-grained local spatial-spectral structures, while the Fourier-based global filtering modules efficiently model long-range dependencies and suppress noise. To further mitigate the severe class imbalance commonly observed in HSIC, HGFNet incorporates an Adaptive Focal Loss (AFL) that dynamically adjusts class-wise focusing and weighting, improving discrimination for underrepresented classes.
Sign in to start a discussion.
More Papers Like This
An Attention-Based 3D Convolutional Autoencoder for Few-Shot Hyperspectral Unmixing and Classification
Researchers developed a semi-supervised deep learning pipeline for hyperspectral image classification that uses a 3D convolutional autoencoder with attention mechanisms to extract spectral-spatial features, achieving better classification accuracy than supervised CNN methods when only small amounts of labeled data are available.
Multiscale Dense Cross-Attention Mechanism with Covariance Pooling for Hyperspectral Image Scene Classification
Researchers developed a multiscale dense cross-attention mechanism with covariance pooling for hyperspectral image scene classification, addressing challenges of high dimensionality and feature redundancy in deep convolutional frameworks to improve classification accuracy.
A new band selection framework for hyperspectral remote sensing image classification
Researchers developed a new framework for reducing data complexity in hyperspectral satellite images by combining dual band selection with a convolutional neural network, achieving over 97% classification accuracy across three benchmark datasets. This approach could improve remote sensing applications like land cover mapping and environmental monitoring.
Deep Kernel Extreme-Learning Machine for the Spectral–Spatial Classification of Hyperspectral Imagery
This paper describes a deep neural network method combining kernel extreme-learning machines with spectral-spatial analysis for classifying hyperspectral remote sensing images. Hyperspectral imaging is also being developed as a tool for detecting and identifying microplastics in environmental samples.
Deep Intrinsic Decomposition with Adversarial Learning for Hyperspectral Image Classification
This paper presents a deep learning method for hyperspectral image classification that accounts for complex environmental variation causing within-class spectral differences. Such techniques may have applications in automated detection and identification of microplastics in environmental samples using spectral imaging.