IEEE Geoscience and Remote Sensing Magazine - December 2021 - 61

with the nonlinearity of HSIs through the kernel technique.
It should be noted that the kernel approach is, in
essence, a template-based model, which results in a large
computing amount and can only alleviate the nonlinear
separable problem to a certain extent, which restricts the
technique's practical applications.
Many deep learning-based methods have been developed
in the computer vision field and shown powerful capabilities
for nonlinear fitting and feature extraction, but the
successful deployments in hyperspectral clustering are very
rare. Generally speaking, current hyperspectral clustering
methods remain at the stage of shallow learning and only
utilize the low-level features of HSIs, which yields a limited
clustering accuracy. Due to the huge differences between
hyperspectral data and natural figures, directly introducing
deep models in the computer vision field to HSIs generally
fails to obtain a satisfactory effect. Therefore, how to adjust/
modify deep models to better learn the intrinsic structure
of HSIs and extract more informatic and discriminative features
to further improve the clustering performance would
be a very promising research line.
AUTOMATICALLY ESTIMATING
THE NUMBER OF CLUSTERS
Automatically and accurately estimating the number of clusters
is very important for hyperspectral clustering, which
promotes clustering applications to be more intelligent and
attractive in practical applications. However, most current
studies focus on improving the clustering models, and studies
on the automatic estimation of the number of clusters are
relatively few. Although some methods can automatically
estimate the number of clusters for HSIs, they are generally
bound to specific clustering models, e.g., FCM, and have
a limited universality. Hence, finding a technique to automatically
and accurately estimate the number of clusters in
a more generic way will be an interesting and important research
direction in the future.
ACKNOWLEDGMENTS
This work was funded, in part, by the Special Foundation
for National Science and Technology Basic Research Program
of China, under grant 2019FY202503; the National
Key Research and Development Program of China, under
grant 2018YFB0504500; the National Natural Science
Foundation of China, under grants 42001313, 61871298,
and 42071322; and the Fundamental Research Funds for
Central Universities, under grant G1323520273. Readers
who have questions about the article are encouraged to directly
contact the corresponding author, Hongyan Zhang
(zhanghongyan@whu.edu.cn).
AUTHOR INFORMATION
Han Zhai (zhaihan@cug.edu.cn) is with the School of
Geography and Information Engineering, China University
of Geosciences, Wuhan, 430074, China. He is a
Member of IEEE.
DECEMBER 2021 IEEE GEOSCIENCE AND REMOTE SENSING MAGAZINE
Hongyan Zhang (zhanghongyan@whu.edu.cn) is with
the State Key Laboratory of Information Engineering in Surveying,
Mapping, and Remote Sensing, Wuhan University,
Wuhan, 430079, China. He is a Senior Member of IEEE.
Pingxiang Li (pxLi@whu.edu.cn) is with the State Key
Laboratory of Information Engineering in Surveying, Mapping,
and Remote Sensing, Wuhan University, Wuhan,
430079, China. He is a Member of IEEE.
Liangpei Zhang (zlp62@whu.edu.cn) is with the State
Key Laboratory of Information Engineering in Surveying,
Mapping, and Remote Sensing, Wuhan University, Wuhan,
430079, China. He is a Fellow of IEEE.
REFERENCES
[1] A. Plaza et al., " Recent advances in techniques for hyperspectral
image processing, " Remote Sens. Environ., vol. 113, pp. S110-
S122, Sept. 2009. doi: 10.1016/j.rse.2007.07.028.
[2] G. Camps-Valls, D. Tuia, L. Bruzzone, and J. A. Benediktsson,
" Advances in hyperspectral image classification: Earth
monitoring with statistical learning methods, " IEEE Signal Process.
Mag., vol. 31, no. 1, pp. 45-54, Jan. 2014. doi: 10.1109/
MSP.2013.2279179.
[3] M. Imani and H. Ghassemian, " An overview on spectral and
spatial information fusion for hyperspectral image classification:
Current trends and challenges, " Inf. Fusion, vol. 59, pp.
59-83, July 2020. doi: 10.1016/j.inffus.2020.01.007.
[4] P. Duan, X. Kang, S. Li, P. Ghamisi, and J. A. Benediktsson, " Fusion
of multiple edge-preserving operations for hyperspectral image
classification, " IEEE Trans. Geosci. Remote Sens., vol. 57, no. 12,
pp. 10,336-10,349, 2019. doi: 10.1109/TGRS.2019.2933588.
[5] H. Zhang, Y. Song, C. Han, and L. Zhang, " Remote sensing image
spatiotemporal fusion using a generative adversarial network, "
IEEE Trans. Geosci. Remote Sens., early access, 2020. doi:
10.1109/TGRS.2020.3010530.
[6] G. Camps-Valls and L. Bruzzone, " Kernel-based methods for
hyperspectral
image classification, " IEEE Trans. Geosci. Remote
Sens., vol. 43, no. 6, pp. 1351-1362, 2005. doi: 10.1109/
TGRS.2005.846154.
[7] H. Zhang, L. Liu, W. He, and L. Zhang, " Hyperspectral image
denoising with total variation regularization and nonlocal
low-rank tensor decomposition, " IEEE Trans. Geosci. Remote
Sens., vol. 58, no. 5, pp. 3071-3084, 2019. doi: 10.1109/
TGRS.2019.2947333.
[8] H. Zhai, H. Zhang, L. Zhang, and P. Li, " Cloud/shadow detection
based on spectral indices for multi/hyperspectral
optical remote sensing imagery, " ISPRS J. Photogram. Remote
Sens., vol. 144, pp. 235-253, Oct. 2018. doi: 10.1016/j.isprsjprs.2018.07.006.
[9]
W. He, H. Zhang, and L. Zhang, " Total variation regularized reweighted
sparse nonegative matrix factorization for hyperspectral
unmixing, " IEEE Trans. Geosci. Remote Sens., vol. 55, no. 7,
pp. 3909-3921, 2017. doi: 10.1109/TGRS.2017.2683719.
[10] F. A. Kruse, J. W. Boardman, and J. F. Huntington, " Comparison
of airborne hyperspectral data and EO-1 Hyperion for mineral
mapping, " IEEE Trans. Geosci. Remote Sens., vol. 41, no. 6,
pp. 1388-1400, 2003. doi: 10.1109/TGRS.2003.812908.
61
IEEE Geoscience and Remote Sensing Magazine - December 2021

Table of Contents for the Digital Edition of IEEE Geoscience and Remote Sensing Magazine - December 2021
