Article
Environmental Sciences
Ying Cheng, Liaoying Zhao, Shuhan Chen, Xiaorun Li
Summary: Spectral unmixing is an important topic in hyperspectral image analysis as it deals with the presence of multiple sources in images and the variability in spectral signatures caused by environmental conditions. Various spectral mixing models have been proposed, but their interpretation is often insufficient and the corresponding unmixing algorithms are classical techniques. This paper introduces a spectral unmixing network based on a scaled and perturbed linear mixing model, incorporating deep learning techniques for determining abundances, scales, and perturbations. The proposed approach outperforms other competitors in both synthetic and real data sets.
Article
Engineering, Electrical & Electronic
Ricardo Augusto Borsoi, Tales Imbiriba, Jose Carlos Moreira Bermudez, Cedric Richard
Summary: The article proposes an efficient multitemporal SU method that utilizes the high temporal correlation between the abundances to provide more accurate results at a lower computational complexity. By separately addressing the endmember selection and the abundance estimation problems, a simpler solution without sacrificing accuracy is achieved. A strategy to detect and address abrupt abundance variations in time is also proposed.
IEEE TRANSACTIONS ON COMPUTATIONAL IMAGING
(2021)
Article
Chemistry, Multidisciplinary
Baohua Jin, Yunfei Zhu, Wei Huang, Qiqiang Chen, Sijia Li
Summary: The purpose of hyperspectral unmixing is to obtain the spectral features and proportions of materials in a hyperspectral image. However, spectral variabilities make it difficult to accurately extract these features. To address this issue, this study proposes an efficient attention-based convolutional neural network and a convolution block attention module. Experimental results demonstrate that this method outperforms other unmixing methods.
APPLIED SCIENCES-BASEL
(2022)
Article
Geochemistry & Geophysics
Behnood Rasti, Bikram Koirala, Paul Scheunders, Jocelyn Chanussot
Summary: This article proposes a minimum simplex convolutional network (MiSiCNet) for deep hyperspectral unmixing. The network incorporates both spatial and geometrical information, leading to robust performance in the absence of pure pixels. Experimental results on simulated and real datasets demonstrate the superiority of MiSiCNet over state-of-the-art unmixing approaches.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2022)
Article
Mathematics, Applied
Yingying Xu, Lei Du, Haifeng Song, Songsong Dai
Summary: The significance of hyperspectral unmixing lies in separating mixed spectral signatures into endmembers and calculating their fractional abundances. However, blind unmixing based on non-negative matrix factorization (NMF) often yields inaccurate and unstable results, while semi-blind unmixing based on sparse regression requires an overcomplete spectral library. To overcome these limitations, a cooperative model called CoNMFSL is proposed, which combines NMF and spectral library to utilize partial prior information from the incomplete spectral library. Extensive simulations and real-data experiments demonstrate that CoNMFSL achieves more accurate and stable results compared to blind unmixing, and relaxes the application conditions of semi-blind unmixing.
JOURNAL OF NONLINEAR AND CONVEX ANALYSIS
(2022)
Article
Environmental Sciences
Ke Wu, Tao Chen, Ying Xu, Dongwei Song, Haishan Li
Summary: The study proposes a novel change detection approach which tackles the challenges of mixed pixels and variations in endmembers by utilizing image stacking and dividing based on spectral unmixing, resulting in improved detection accuracy. Experimental results show that in simulated data, the overall accuracy and Kappa coefficient values reached 99.61% and 0.99. For two sets of real data, the highest accuracy rates were 93.26% and 80.85%, showing an increase of 14.88% and 13.42% compared to the worst results. Additionally, the Kappa coefficient values were consistent with the accuracy rates.
Article
Chemistry, Analytical
Salah Eddine Brezini, Yannick Deville
Summary: The aim of fusing hyperspectral and multispectral images is to improve their spatial resolutions. However, most fusion methods do not consider spectral variability, leading to reduced quality of the sharpened products. This paper introduces a new method called HSB-SV, which addresses spectral variability using spectra bundles-based method and sparsity-based method.
Article
Computer Science, Artificial Intelligence
Jin-Ju Wang, Ding-Cheng Wang, Ting-Zhu Huang, Jie Huang, Xi-Le Zhao, Liang-Jian Deng
Summary: In this paper, a new NTF-based model called EIC-NTF is proposed for hyperspectral unmixing to mitigate the impact of high correlation among endmembers and abundances. The model introduces endmember independence constraint for endmember estimation and exploits the low-rankness in abundance maps for abundance estimation. Experimental results show that the proposed algorithm is effective for hyperspectral unmixing.
KNOWLEDGE-BASED SYSTEMS
(2021)
Article
Geochemistry & Geophysics
Wendi Liu, Xiaoguang Mei, Yong Ma, Jun Huang, Qihai Chen, Hao Li
Summary: In this letter, a new hyperspectral unmixing algorithm based on graph Laplacian regularization and l(1)-norm-Gaussian mixture model is proposed. The algorithm effectively addresses the issues of fixed endmembers and underutilized spatial characteristics of the hyperspectral image. Experimental results demonstrate its superiority over other state-of-the-art methods.
IEEE GEOSCIENCE AND REMOTE SENSING LETTERS
(2023)
Article
Environmental Sciences
Jingyan Zhang, Xiangrong Zhang, Licheng Jiao
Summary: In this paper, a sparse NMF algorithm based on endmember independence and spatial weighted abundance is proposed to address the issue of inaccurate endmember extraction in hyperspectral unmixing. The algorithm considers both the relevant characteristics of endmembers and abundances simultaneously, and makes full use of the spatial-spectral information in the image, achieving a more desired unmixing performance.
Article
Geochemistry & Geophysics
Dengyong Zhang, Taowei Wang, Shujun Yang, Yuheng Jia, Feng Li
Summary: Sparse unmixing separates the pixels of hyperspectral images into pure spectral signatures and coefficients, avoiding the inaccuracy of endmember extraction. The fast multiscale spatial regularization unmixing algorithm ignores pixel correlation and spectral variability, which we address by introducing weighting factors to improve the unmixing result.
IEEE GEOSCIENCE AND REMOTE SENSING LETTERS
(2022)
Article
Geochemistry & Geophysics
Hongyi Liu, Youkang Lu, Zebin Wu, Qian Du, Jocelyn Chanussot, Zhihui Wei
Summary: This article proposes a method for unmixing hyperspectral image sequences in the wavelet domain using the Bayesian method. By representing the intrinsic and invariant features of the spectral curve through wavelet transform and solving the model using the Markov chain Monte Carlo sampling algorithm, more accurate results are obtained in the experiments.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2023)
Article
Environmental Sciences
Ziqiang Hua, Xiaorun Li, Jianfeng Jiang, Liaoying Zhao
Summary: This paper proposes two gated autoencoder networks that adaptively control the contribution of spectral and spatial features in spectral unmixing, filtering and regularization spatial features through gating mechanism, achieving superior experimental results compared to state-of-the-art techniques.
Article
Geochemistry & Geophysics
Taner Ince, Nicolas Dobigeon
Summary: In this paper, we propose a simple yet efficient sparse unmixing method for hyperspectral images, which exploits the spatial and spectral properties of the images. The proposed method performs a sparse unmixing on a coarse hyperspectral image obtained through spatial smoothing, and then uses the estimated coarse abundance map to design a sparse regularization for the original hyperspectral image. Experimental results demonstrate that the proposed method achieves competitive performance with lower computational complexity compared to state-of-the-art methods.
IEEE GEOSCIENCE AND REMOTE SENSING LETTERS
(2023)
Article
Engineering, Electrical & Electronic
Saeideh Ghanbari Azar, Saeed Meshgini, Soosan Beheshti, Tohid Yousefi Rezaii
Summary: This paper introduces the process of hyperspectral unmixing and methods to address spectral variability, proposing the LMM-SBD method that combines scaling factors and bundle dictionary. By considering spatial coherence of neighboring pixels, the method successfully tackles the unmixing problem.
