Hyperspectral unmixing based on adversarial autoencoder network
- Vol. 27, Issue 8, Pages: 1964-1974(2023)
Published: 07 August 2023
DOI: 10.11834/jrs.20210550
扫 描 看 全 文
浏览全部资源
扫码关注微信
Published: 07 August 2023 ,
扫 描 看 全 文
靳淇文,马泳,樊凡,黄珺,李皞,梅晓光.2023.生成对抗网络的无监督高光谱解混.遥感学报,27(8): 1964-1974
Jin Q W,Ma Y,Fan F,Huang J,Li H and Mei X G. 2023. Hyperspectral unmixing based on adversarial autoencoder network. National Remote Sensing Bulletin, 27(8):1964-1974
近年来,基于深度学习中自编码器(AE)的方法在无监督的高光谱解混中受到了广泛的关注。由于AE的学习过程可以表述为通过训练找出一组低维的隐藏层(丰度),并用其对应的权重(端元)进行组合来减少重构误差,因而这种框架被广泛迁移并应用于高光谱的解混算法之中。然而,现有基于AE的框架虽然能有效地处理无监督的解混场景,却都存在着对噪声和初始化条件不鲁棒的问题,且解混精度也有待进一步提升。针对以上问题本文提出了一种全新的基于对抗性自编码网络(AAE)的无监督解混网络框架。首先,在网络的生成器中根据丰度和为一(ASC)及非负性(ANC)的物理意义,设计了一个基于AE的端到端解混框架。然后,在网络的判决器中本文采用初始化的丰度图作为真实值,将生成器的隐藏层(丰度)与初始化的丰度进行对抗训练,在重构误差与对抗误差的同步优化中提升框架解混性能。与传统的AE方法相比,该方法通过引入对抗性过程,在判决器中加入丰度的先验知识,可以大大提高框架的性能和鲁棒性。仿真和真实的高光谱数据的实验表明,该算法较现有方法相比具有更高的解混精度。
Due to the limitations associated with the spatial resolution of instruments and complex natural surfaces
spectral mixing (SU)
which identifies the proportion (abundance) of the basic component spectrum (endmember) in the sub-pixel level
has become an important topic in the deep development of hyperspectral image analysis. Given that the training procedure can be explained as finding a set of low-dimensional representations (abundance) that reconstruct the data with their corresponding bases (endmembers)
the autoencoder (AE)-based method has received much attention in unsupervised hyperspectral unmixing. However
although the existing AE methods can effectively deal with unsupervised unmixing scenarios
their performance has not been satisfactory
and noise and initialization conditions greatly affect their unmixing performance.
In this paper
we propose AAENet
a novel network for unsupervised unmixing that is based on the adversarial autoencoder network. First
we design the generator as an end-to-end unmixing network based on AE to obtain the meaningful abundance subjected to Abundance Nonnegative Constraint (ANC) and Abundance Sum-to-one Constraint (ASC). Second
we take the adversarial training process to map the abundance prior to the hidden code vector (abundance)
which is equivalent to providing an adaptive training error to correct the AAENet converging toward a highly accurate and interpretable unmixing solution.
Experiments on simulated and real hyperspectral data (Jasper dataset) demonstrate that the proposed algorithm can outperform the state-of-the-art methods. The synthetic data are polluted by Gaussian noise at different levels
where the SNR varies from 10 dB to 30 dB with an interval of 10 dB. Each algorithm is run 10 times
and the average and standard deviation are reported. With an increasing noise level
the proposed algorithm exhibits higher robustness in both the abundance and endmember estimations and achieves the best or comparable results in all cases. In experiments on real datasets
AAENet not only shows sparse abundances for the region but also interprets the boundary as a combination of neighboring materials. The best results are obtained in highly mixed scenes.
Compared with the traditional AE method
the proposed algorithm can greatly enhance the performance and robustness of the model by using the adversarial procedure and incorporating the abundance prior to the framework. The discrimination network is designed to allow the transfer of the potentially intrinsic properties of the abundance prior information. As its main purpose
the proposed method takes an adversarial training process to impose a prior on the hidden code vector of the autoencoder. The output of the hidden code vector is then guided to produce meaningful samples. Experiments on simulated and real hyperspectral data demonstrate that the proposed algorithm can achieve a better unmixing performance compared with state-of-the-art methods.
遥感高光谱解混深度学习对抗自编码器高光谱图像
remote sensinghyperspectral unmixingdeep learningadversarial autoencoderhyperspectral image
Bioucas-Dias J M, Plaza A, Dobigeon N, Parente M, Du Q, Gader P and Chanussot J. 2012. Hyperspectral unmixing overview: geometrical, statistical, and sparse regression-based approaches. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 5(2): 354-379 [DOI: 10.1109/JSTARS.2012.2194696http://dx.doi.org/10.1109/JSTARS.2012.2194696]
Borsoi R A, Imbiriba T and Bermudez J C M. 2020. Deep generative endmember modeling: an application to unsupervised spectral unmixing. IEEE Transactions on Computational Imaging, 6: 374-384 [DOI: 10.1109/tci.2019.2948726http://dx.doi.org/10.1109/tci.2019.2948726]
Eches O, Dobigeon N, Mailhes C and Tourneret J Y. 2010. Bayesian estimation of linear mixtures using the normal compositional model. Application to hyperspectral imagery. IEEE Transactions on Image Processing, 19(6): 1403-1413 [DOI: 10.1109/TIP.2010.2042993http://dx.doi.org/10.1109/TIP.2010.2042993]
Goodfellow I J, Pouget-Abadie J, Mirza M, Xu B, Warde-Farley D, Ozair S, Courville A and Bengio Y. 2014. Generative adversarial nets//Proceedings of the 27th International Conference on Neural Information Processing Systems. Montreal: MIT Press: 2672-2680.
Han Z, Gao L R, Zhang B, Sun X and Li Q T. 2020. Nonlinear hyperspectral unmixing algorithm based on deep autoencoder networks. Journal of Remote Sensing, 24(4): 388-400
韩竹, 高连如, 张兵, 孙旭, 李庆亭. 2020. 高分五号高光谱图像自编码网络非线性解混. 遥感学报, 24(4): 388-400 [DOI: 10.11834/jrs.20209188http://dx.doi.org/10.11834/jrs.20209188]
Heinz D C and Chein-I-Chang. 2001. Fully constrained least squares linear spectral mixture analysis method for material quantification in hyperspectral imagery. IEEE Transactions on Geoscience and Remote Sensing, 39(3): 529-545 [DOI: 10.1109/36.911111http://dx.doi.org/10.1109/36.911111]
Jin Q W, Ma Y, Mei X G, Dai X B, Li H, Fan F and Huang J. 2019a. Gaussian mixture model for hyperspectral unmixing with low-rank representation//Proceedings of 2019 IEEE International Geoscience and Remote Sensing Symposium. Yokohama: IEEE: 294-297 [DOI: 10.1109/IGARSS.2019.8898410http://dx.doi.org/10.1109/IGARSS.2019.8898410]
Jin Q W, Ma Y, Mei X G, Li H and Ma J Y. 2021. UTDN: an unsupervised two-stream dirichlet-net for hyperspectral unmixing//Proceedings of 2021 IEEE International Conference on Acoustics, Speech and Signal Processing. Toronto: IEEE: 1885-1889 [DOI: 10.1109/ICASSP39728.2021.9414810http://dx.doi.org/10.1109/ICASSP39728.2021.9414810]
Jin Q W, Ma Y, Pan E T, Fan F, Huang J, Li H, Sui C H and Mei X G. 2019b. Hyperspectral unmixing with gaussian mixture model and spatial group sparsity. Remote Sensing, 11(20): 2434 [DOI: 10.3390/rs11202434http://dx.doi.org/10.3390/rs11202434]
Ma Y, Fan G H, Jin Q W, Huang J, Mei X G and Ma J Y. 2021. Hyperspectral anomaly detection via integration of feature extraction and background purification. IEEE Geoscience and Remote Sensing Letters, 18(8): 1436-1440 [DOI: 10.1109/LGRS.2020.2998809http://dx.doi.org/10.1109/LGRS.2020.2998809]
Ma Y, Jin Q W, Mei X G, Dai X B, Fan F, Li H and Huang J. 2019. Hyperspectral unmixing with Gaussian mixture model and low-rank representation. Remote Sensing, 11(8): 911 [DOI: 10.3390/rs11080911http://dx.doi.org/10.3390/rs11080911]
Mei X G, Ma Y, Li C, Fan F, Huang J and Ma J Y. 2018. Robust GBM hyperspectral image unmixing with superpixel segmentation based low rank and sparse representation. Neurocomputing, 275: 2783-2797 [DOI: 10.1016/j.neucom.2017.11.052http://dx.doi.org/10.1016/j.neucom.2017.11.052]
Palsson B, Ulfarsson M O and Sveinsson J R. 2021. Convolutional autoencoder for spectral-spatial hyperspectral unmixing. IEEE Transactions on Geoscience and Remote Sensing, 59(1): 535-549 [DOI: 10.1109/tgrs.2020.2992743http://dx.doi.org/10.1109/tgrs.2020.2992743]
Su Y C, Li J, Plaza A, Marinoni A, Gamba P and Chakravortty S. 2019. DAEN: deep autoencoder networks for hyperspectral unmixing. IEEE Transactions on Geoscience and Remote Sensing, 57(7): 4309-4321 [DOI: 10.1109/tgrs.2018.2890633http://dx.doi.org/10.1109/tgrs.2018.2890633]
Tong Q X, Zhang B and Zhang L F. 2016. Current progress of hyperspectral remote sensing in China. Journal of Remote Sensing, 20(5): 689-707
童庆禧, 张兵, 张立福. 2016. 中国高光谱遥感的前沿进展. 遥感学报, 20(5): 689-707 [DOI: 10.11834/jrs.20166264http://dx.doi.org/10.11834/jrs.20166264]
Ying Q and Qi H R. 2019. uDAS: an untied denoising autoencoder with sparsity for spectral unmixing. IEEE Transactions on Geoscience and Remote Sensing, 57(3): 1698-1712 [DOI: 10.1109/tgrs.2018.2868690http://dx.doi.org/10.1109/tgrs.2018.2868690]
Zhou Y, Rangarajan A and Gader P D. 2016. A spatial compositional model for linear unmixing and endmember uncertainty estimation. IEEE Transactions on Image Processing, 25(12): 5987-6002 [DOI: 10.1109/tip.2016.2618002http://dx.doi.org/10.1109/tip.2016.2618002]
Zhou Y, Rangarajan A and Gader P D. 2018. A Gaussian mixture model representation of endmember variability in hyperspectral unmixing. IEEE Transactions on Image Processing, 27(5): 2242-2256 [DOI: 10.1109/TIP.2018.2795744http://dx.doi.org/10.1109/TIP.2018.2795744]
Zhu F. Hyperspectral unmixing datasets and ground truths. 2014
相关文章
相关作者
相关机构