Hyperspectral Image Classification Using Spectral-Spatial Dual Random Fields With Gaussian and Markov Processes

This article presents a novel hyperspectral image (HSI) classification approach that integrates the sparse inducing variational Gaussian process (SIVGP) with a spatially adaptive Markov random field (SAMRF), termed G-MDRF. Variational inference is employed to obtain a sparse approximation of the pos...

Full description

Saved in:
Bibliographic Details
Main Authors: Yaqiu Zhang, Lizhi Liu, Xinnian Yang
Format: Article
Language:English
Published: IEEE 2025-01-01
Series:IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
Subjects:
Online Access:https://ieeexplore.ieee.org/document/10836880/
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1832576794973175808
author Yaqiu Zhang
Lizhi Liu
Xinnian Yang
author_facet Yaqiu Zhang
Lizhi Liu
Xinnian Yang
author_sort Yaqiu Zhang
collection DOAJ
description This article presents a novel hyperspectral image (HSI) classification approach that integrates the sparse inducing variational Gaussian process (SIVGP) with a spatially adaptive Markov random field (SAMRF), termed G-MDRF. Variational inference is employed to obtain a sparse approximation of the posterior distribution, modeling the spectral field within the latent function space. Subsequently, SAMRF is utilized to model the spatial prior within the function space, while the alternating direction method of multipliers (ADMM) is employed to enhance computational efficiency. Experimental results on three datasets with varying complexity show that the proposed algorithm improves computational efficiency by approximately 152 times and accuracy by about 7%–26% compared to the current popular Gaussian process methods. Compared to classical random field methods, G-MDRF rapidly achieves a convergent solution with only one ten-thousandth to one hundred-thousandth of the iterations, improving accuracy by about 5%–18%. Particularly, when the number of classes in the dataset increases and the scene becomes more complex, the proposed method demonstrates a greater advantage in both computational efficiency and classification accuracy compared to existing methods.
format Article
id doaj-art-ecd738aeed31400a8e403f3c7438748d
institution Kabale University
issn 1939-1404
2151-1535
language English
publishDate 2025-01-01
publisher IEEE
record_format Article
series IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
spelling doaj-art-ecd738aeed31400a8e403f3c7438748d2025-01-31T00:00:31ZengIEEEIEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing1939-14042151-15352025-01-01184199421210.1109/JSTARS.2025.352811510836880Hyperspectral Image Classification Using Spectral-Spatial Dual Random Fields With Gaussian and Markov ProcessesYaqiu Zhang0https://orcid.org/0000-0002-0819-0799Lizhi Liu1Xinnian Yang2https://orcid.org/0009-0002-9493-0980School of Electrical and Information Engineering, Hubei University of Automotive Technology, Shiyan, ChinaCollege of Landscape and Forestry, Tarim University, Xinjiang, ChinaSchool of Intelligent Connected Vehicle, Hubei University of Automotive Technology, Shiyan, ChinaThis article presents a novel hyperspectral image (HSI) classification approach that integrates the sparse inducing variational Gaussian process (SIVGP) with a spatially adaptive Markov random field (SAMRF), termed G-MDRF. Variational inference is employed to obtain a sparse approximation of the posterior distribution, modeling the spectral field within the latent function space. Subsequently, SAMRF is utilized to model the spatial prior within the function space, while the alternating direction method of multipliers (ADMM) is employed to enhance computational efficiency. Experimental results on three datasets with varying complexity show that the proposed algorithm improves computational efficiency by approximately 152 times and accuracy by about 7%–26% compared to the current popular Gaussian process methods. Compared to classical random field methods, G-MDRF rapidly achieves a convergent solution with only one ten-thousandth to one hundred-thousandth of the iterations, improving accuracy by about 5%–18%. Particularly, when the number of classes in the dataset increases and the scene becomes more complex, the proposed method demonstrates a greater advantage in both computational efficiency and classification accuracy compared to existing methods.https://ieeexplore.ieee.org/document/10836880/Gaussian processes (GPs)hyperspectral image (HSI) classificationMarkov random field (MRF)stochastic processes
spellingShingle Yaqiu Zhang
Lizhi Liu
Xinnian Yang
Hyperspectral Image Classification Using Spectral-Spatial Dual Random Fields With Gaussian and Markov Processes
IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
Gaussian processes (GPs)
hyperspectral image (HSI) classification
Markov random field (MRF)
stochastic processes
title Hyperspectral Image Classification Using Spectral-Spatial Dual Random Fields With Gaussian and Markov Processes
title_full Hyperspectral Image Classification Using Spectral-Spatial Dual Random Fields With Gaussian and Markov Processes
title_fullStr Hyperspectral Image Classification Using Spectral-Spatial Dual Random Fields With Gaussian and Markov Processes
title_full_unstemmed Hyperspectral Image Classification Using Spectral-Spatial Dual Random Fields With Gaussian and Markov Processes
title_short Hyperspectral Image Classification Using Spectral-Spatial Dual Random Fields With Gaussian and Markov Processes
title_sort hyperspectral image classification using spectral spatial dual random fields with gaussian and markov processes
topic Gaussian processes (GPs)
hyperspectral image (HSI) classification
Markov random field (MRF)
stochastic processes
url https://ieeexplore.ieee.org/document/10836880/
work_keys_str_mv AT yaqiuzhang hyperspectralimageclassificationusingspectralspatialdualrandomfieldswithgaussianandmarkovprocesses
AT lizhiliu hyperspectralimageclassificationusingspectralspatialdualrandomfieldswithgaussianandmarkovprocesses
AT xinnianyang hyperspectralimageclassificationusingspectralspatialdualrandomfieldswithgaussianandmarkovprocesses