An Enhanced Spectral Fusion 3D CNN Model for Hyperspectral Image Classification

With the continuous development of hyperspectral image technology and deep learning methods in recent years, an increasing number of hyperspectral image classification models have been proposed. However, due to the numerous spectral dimensions of hyperspectral images, most classification models suff...

Full description

Bibliographic Details
Main Authors: Junbo Zhou, Shan Zeng, Zuyin Xiao, Jinbo Zhou, Hao Li, Zhen Kang
Format: Article
Language:English
Published: MDPI AG 2022-10-01
Series:Remote Sensing
Subjects:
Online Access:https://www.mdpi.com/2072-4292/14/21/5334
_version_ 1827645652885045248
author Junbo Zhou
Shan Zeng
Zuyin Xiao
Jinbo Zhou
Hao Li
Zhen Kang
author_facet Junbo Zhou
Shan Zeng
Zuyin Xiao
Jinbo Zhou
Hao Li
Zhen Kang
author_sort Junbo Zhou
collection DOAJ
description With the continuous development of hyperspectral image technology and deep learning methods in recent years, an increasing number of hyperspectral image classification models have been proposed. However, due to the numerous spectral dimensions of hyperspectral images, most classification models suffer from issues such as breaking spectral continuity and poor learning of spectral information. In this paper, we propose a new classification model called the enhanced spectral fusion network (ESFNet), which contains two parts: an optimized multi-scale fused spectral attention module (FsSE) and a 3D convolutional neural network (3D CNN) based on the fusion of different spectral strides (SSFCNN). Specifically, after sampling the hyperspectral images, our model first implements the weighting of the spectral information through the FsSE module to obtain spectral data with a higher degree of information richness. Then, the weighted spectral data are fed into the SSFCNN to realize the effective learning of spectral features. The new model can maximize the retention of spectral continuity and enhance the spectral information while being able to better utilize the enhanced information to improve the model’s ability to learn hyperspectral image features, thus improving the classification accuracy of the model. Experiment results on the Indian Pines and Pavia University datasets demonstrated that our method outperforms other relevant baselines in terms of classification accuracy and generalization performance.
first_indexed 2024-03-09T18:42:11Z
format Article
id doaj.art-8cad9fa1f67746e59b3a6fc8de08fc7b
institution Directory Open Access Journal
issn 2072-4292
language English
last_indexed 2024-03-09T18:42:11Z
publishDate 2022-10-01
publisher MDPI AG
record_format Article
series Remote Sensing
spelling doaj.art-8cad9fa1f67746e59b3a6fc8de08fc7b2023-11-24T06:37:21ZengMDPI AGRemote Sensing2072-42922022-10-011421533410.3390/rs14215334An Enhanced Spectral Fusion 3D CNN Model for Hyperspectral Image ClassificationJunbo Zhou0Shan Zeng1Zuyin Xiao2Jinbo Zhou3Hao Li4Zhen Kang5School of Mathematics and Computer Science, Wuhan Polytechnic University, Wuhan 430023, ChinaSchool of Mathematics and Computer Science, Wuhan Polytechnic University, Wuhan 430023, ChinaSchool of Mathematics and Computer Science, Wuhan Polytechnic University, Wuhan 430023, ChinaSchool of Mathematics and Computer Science, Wuhan Polytechnic University, Wuhan 430023, ChinaSchool of Mathematics and Computer Science, Wuhan Polytechnic University, Wuhan 430023, ChinaSchool of Mathematics and Computer Science, Wuhan Polytechnic University, Wuhan 430023, ChinaWith the continuous development of hyperspectral image technology and deep learning methods in recent years, an increasing number of hyperspectral image classification models have been proposed. However, due to the numerous spectral dimensions of hyperspectral images, most classification models suffer from issues such as breaking spectral continuity and poor learning of spectral information. In this paper, we propose a new classification model called the enhanced spectral fusion network (ESFNet), which contains two parts: an optimized multi-scale fused spectral attention module (FsSE) and a 3D convolutional neural network (3D CNN) based on the fusion of different spectral strides (SSFCNN). Specifically, after sampling the hyperspectral images, our model first implements the weighting of the spectral information through the FsSE module to obtain spectral data with a higher degree of information richness. Then, the weighted spectral data are fed into the SSFCNN to realize the effective learning of spectral features. The new model can maximize the retention of spectral continuity and enhance the spectral information while being able to better utilize the enhanced information to improve the model’s ability to learn hyperspectral image features, thus improving the classification accuracy of the model. Experiment results on the Indian Pines and Pavia University datasets demonstrated that our method outperforms other relevant baselines in terms of classification accuracy and generalization performance.https://www.mdpi.com/2072-4292/14/21/5334deep learninghyperspectral image classificationattention mechanismfeature fusion3D CNN
spellingShingle Junbo Zhou
Shan Zeng
Zuyin Xiao
Jinbo Zhou
Hao Li
Zhen Kang
An Enhanced Spectral Fusion 3D CNN Model for Hyperspectral Image Classification
Remote Sensing
deep learning
hyperspectral image classification
attention mechanism
feature fusion
3D CNN
title An Enhanced Spectral Fusion 3D CNN Model for Hyperspectral Image Classification
title_full An Enhanced Spectral Fusion 3D CNN Model for Hyperspectral Image Classification
title_fullStr An Enhanced Spectral Fusion 3D CNN Model for Hyperspectral Image Classification
title_full_unstemmed An Enhanced Spectral Fusion 3D CNN Model for Hyperspectral Image Classification
title_short An Enhanced Spectral Fusion 3D CNN Model for Hyperspectral Image Classification
title_sort enhanced spectral fusion 3d cnn model for hyperspectral image classification
topic deep learning
hyperspectral image classification
attention mechanism
feature fusion
3D CNN
url https://www.mdpi.com/2072-4292/14/21/5334
work_keys_str_mv AT junbozhou anenhancedspectralfusion3dcnnmodelforhyperspectralimageclassification
AT shanzeng anenhancedspectralfusion3dcnnmodelforhyperspectralimageclassification
AT zuyinxiao anenhancedspectralfusion3dcnnmodelforhyperspectralimageclassification
AT jinbozhou anenhancedspectralfusion3dcnnmodelforhyperspectralimageclassification
AT haoli anenhancedspectralfusion3dcnnmodelforhyperspectralimageclassification
AT zhenkang anenhancedspectralfusion3dcnnmodelforhyperspectralimageclassification
AT junbozhou enhancedspectralfusion3dcnnmodelforhyperspectralimageclassification
AT shanzeng enhancedspectralfusion3dcnnmodelforhyperspectralimageclassification
AT zuyinxiao enhancedspectralfusion3dcnnmodelforhyperspectralimageclassification
AT jinbozhou enhancedspectralfusion3dcnnmodelforhyperspectralimageclassification
AT haoli enhancedspectralfusion3dcnnmodelforhyperspectralimageclassification
AT zhenkang enhancedspectralfusion3dcnnmodelforhyperspectralimageclassification