Remote Sensing for Natural Resources >
Hyperspectral unmixing based on spectral-spatial total variation nonnegative matrix factorization with feature space augmentation
Received date: 2024-04-01
Revised date: 2024-07-07
Online published: 2026-06-03
Nonnegative matrix factorization (NMF) is commonly used in hyperspectral image (HSI) unmixing due to its high interpretability and computability. To effectively address HSI noise and improve unmixing efficiency, this study proposed a method for hyperspectral unmixing based on spectral-spatial total variation nonnegative matrix factorization (SSTVNMF) with feature space augmentation. First, the original data space was converted to the feature space through feature extraction, allowing the unmixing process to be performed in the feature space for enhanced unmixing efficiency. Second, to reduce the impact of noise, the spatial information was extracted using the bilateral filtering (BF) method for enhanced feature extraction, thereby ensuring the accuracy of extracted features. Third, to ensure the effectiveness of the unmixing method, total variation (TV) regularization that considers both spatial and spectral features was established based on the NMF method. The spatial TV promotes abundance smoothing by calculating the horizontal and vertical differences in abundance between neighboring pixels. Based on the minimum-volume TV, the spectral TV enhances endmember extraction by applying constraint forces between endmembers to minimize the volume. Finally, the proposed method was verified using the synthetic data from the USGS spectral library as simulated data and the Jasper Ridge, APEX, and Cuprite datasets as actual data. The experimental results demonstrate that the proposed method outperformed other improved NMF-based methods in terms of qualitative and quantitative assessments.
QIN Ziyi , YANG Longshan . Hyperspectral unmixing based on spectral-spatial total variation nonnegative matrix factorization with feature space augmentation[J]. Remote Sensing for Natural Resources, 2025 , 37(4) : 21 -30 . DOI: 10.6046/zrzyyg.2024116
图5 特征空间增强下的SSTVNMF与其他对比方法在Jasper Ridge数据集上的端元提取图Fig.5 Endmember extraction of SSTVNMF in feature space and other comparison methods on the Jasper Ridge dataset |
表1 特征空间增强下的SSTVNMF与其他对比方法在Jasper Ridge数据集上的丰度估计Tab.1 Abundance estimation of SSTVNMF in feature space and other comparison methods on the Jasper Ridge dataset |
![]() |
表2 特征空间增强下的SSTVNMF与其他对比方法在3个真实数据集上的评价指标和运行时间Tab.2 Evaluation metrics and running time of SSTVNMF in feature space and other comparison methods on true dataset |
| 算法 | Jasper Ridge数据集 | APEX数据集 | Cuprite数据集 | ||||||
|---|---|---|---|---|---|---|---|---|---|
| SAD | SRE | 运行时间/s | SAD | SRE | 运行时间/s | SAD | SRE | 运行时间/s | |
| 本文方法 | 0.141 6 | 10.590 9 | 0.51 | 0.138 0 | 11.762 0 | 4.23 | 0.398 4 | 9.284 1 | 0.76 |
| SSTVNMF | 0.141 9 | 10.505 2 | 0.32 | 0.138 1 | 11.761 9 | 3.78 | 0.391 7 | 9.177 2 | 0.50 |
| NMFQMV | 0.147 9 | 8.976 2 | 1.20 | 0.138 3 | 11.963 5 | 1.08 | 0.400 0 | 8.272 6 | 1.31 |
| SeCoDe | 0.215 8 | 8.420 6 | 22.78 | 0.163 5 | 3.079 8 | 6.20 | 0.280 5 | 7.026 8 | 13.50 |
| gtvMBO | 0.217 9 | 8.073 6 | 0.01 | 0.141 1 | -0.274 9 | 0.01 | 0.263 7 | 4.308 5 | 0.01 |
| PISINMF | 0.119 8 | 9.916 9 | 2.70 | 0.147 9 | 7.380 3 | 1.53 | 0.434 3 | 10.044 9 | 3.70 |
| EBEAE-TV | 0.237 7 | 11.701 6 | 2.13 | 0.129 8 | -1.452 3 | 1.17 | 0.288 2 | 4.799 4 | 3.37 |
| SGSNMF | 0.257 7 | 7.170 6 | 3.93 | 0.146 0 | 3.185 9 | 0.71 | 0.274 8 | 5.097 6 | 6.56 |
| KbSNMF | 0.137 8 | 9.243 5 | 6.83 | 0.139 8 | 11.132 1 | 3.05 | 0.409 2 | 8.793 3 | 14.74 |
表3 在Jasper Ridge数据集的光谱TV、空间TV和双边滤波消融实验 Fig.3 Blation experiments with Jasper Ridge data for TVspec, TVspa, and BF |
| 算法 | 约束项 | 评价指标 | |||
|---|---|---|---|---|---|
| 空间TV | 光谱TV | BF | SAD | SRE | |
| NMFQMV | √① | × | × | 0.147 9 | 8.976 2 |
| SSTVNMF | √ | √ | × | 0.141 9 | 10.505 2 |
| 本文方法 | √ | √ | √ | 0.141 6 | 10.590 9 |
①√代表加入该约束项,×代表未加入该约束项。 |
| [1] |
苏远超, 许若晴, 高连如, 等. 基于深度学习的高光谱遥感图像混合像元分解研究综述[J]. 遥感学报, 2024, 28(1):1-19.
|
| [2] |
孙肖, 徐林林, 王晓阳, 等. 基于优化K-P-Means解混方法的高光谱图像矿物识别[J]. 自然资源遥感, 2022, 34(3):43-49.doi:10.6046/zrzyyg.2021215.
|
| [3] |
袁静, 章毓晋, 高方平. 线性高光谱解混模型综述[J]. 红外与毫米波学报, 2018, 37(5):553-571.
|
| [4] |
李殷娜, 李正强, 郑杨, 等. 基于非负矩阵分解的中红外地表特性光谱重建方法[J]. 光谱学与光谱分析, 2024, 44(2):563-570.
|
| [5] |
|
| [6] |
王建宇, 李春来. 高光谱遥感成像技术的发展与展望[J]. 空间科学学报, 2021, 41(1):22-33.
|
| [7] |
|
| [8] |
|
| [9] |
|
| [10] |
|
| [11] |
|
| [12] |
|
| [13] |
|
| [14] |
|
| [15] |
|
| [16] |
|
| [17] |
|
| [18] |
|
| [19] |
|
| [20] |
|
| [21] |
|
/
| 〈 |
|
〉 |