Hyperspectral Image Classification Based on Two-Branch Spectral–Spatial-Feature Attention Network

Abstract

Although most of deep-learning-based hyperspectral image (HSI) classification methods achieve great performance, there still remains a challenge to utilize small-size training samples to remarkably enhance the classification accuracy. To tackle this challenge, a novel two-branch spectral–spatial-feature attention network (TSSFAN) for HSI classification is proposed in this paper. Firstly, two inputs with different spectral dimensions and spatial sizes are constructed, which can not only reduce the redundancy of the original dataset but also accurately explore the spectral and spatial features. Then, we design two parallel 3DCNN branches with attention modules, in which one focuses on extracting spectral features and adaptively learning the more discriminative spectral channels, and the other focuses on exploring spatial features and adaptively learning the more discriminative spatial structures. Next, the feature attention module is constructed to automatically adjust the weights of different features based on their contributions for classification to remarkably improve the classification performance. Finally, we design the hybrid architecture of 3D–2DCNN to acquire the final classification result, which can significantly decrease the sophistication of the network. Experimental results on three HSI datasets indicate that our presented TSSFAN method outperforms several of the most advanced classification methods.

Non-destructive detection of single-seed viability in maize using hyperspectral imaging technology and multi-scale 3D convolutional neural network

“…After 400 iterations, the accuracy of the training set remained at around 100%, while the accuracy of the test set remained at around 90% (Figures 6A, C). By using 3DCNN to process the data, not only the spectral information was considered (Wu et al, 2021), but also the image information was integrated, making the evaluation of maize seed quality more comprehensive and accurate (Collins et al, 2021). Compared with machine learning methods that using all spectral bands as input data, the 3DCNN method only used few representative bands.…”

Atmospheric correction of vegetation reflectance with simulation-trained deep learning for ground-based hyperspectral remote sensing

“…Convolutional neural networks are prevalent in the field of hyperspectral image analysis due to their effectiveness at tasks such as image segmentation [68][69][70] and pixel classification [71][72][73]. In recent years, CNNs have been utilized to develop deep learning models to address the problem of extracting the spectral reflectance from the spectral radiance.…”

Leveraging Hyperspectral Images for Accurate Insect Classification with a Novel Two-Branch Self-Correlation Approach

“…Machine learning tools including Support Vector Machines (SVM) [21], K Nearest Neighbors [22], Random Forests [23], and Logistic Regression [24], have shown efficacy in hyperspectral image classification. Concurrently, deep learning-based hyperspectral classification methods [25][26][27][28] are emerging as research trends, offering new insights and techniques for spectral data processing and analysis. Based on the above content, researchers are able to distinguish objects that appear the same in traditional RGB images based on rich spectral information, which can be used for non-destructive chemical and biological analysis.…”