Deep 2D Convolutional Neural Network Architecture for Hyperspectral Land Cover Classification: A Comparative Study with KNN
Abstract
In recent years, deep learning techniques have received a great deal of attention in the context of hyperspectralimage (HSI) classification, particularly with regard to land cover mapping. Although 2D convolutionalneural networks (CNNs) are now widely used in this field, this study presents a refined, deeply structured2D-CNN architecture that is specifically designed for spatial–spectral integration. Rather than introducinga novel concept, the contribution lies in the balanced design of the architecture, which integrates dropoutand batch normalisation to enhance accuracy and generalisability on benchmark datasets. The proposednetwork includes 10 convolutional layers organized into three blocks, each followed by max-pooling, batchnormalization, and dropout layers to reduce overfitting and improve model robustness. A fully connectedclassifier with Softmax activation performs the final prediction. We trained the architecture using the SalinasValley dataset, which contains 54,129 labeled pixels across 16 land cover classes. The data weremeticulously segmented into two distinct components: the initial segment encompassed the primary dataset, while the subsequent segment comprised the ensuing data. It is noteworthy that 70% of the data wasallocated for training purposes. The remaining 30% of the budget was allocated for testing purposes. Thetraining was executed for 100 epochs by employing the Adam optimizer and categorical cross-entropy lossfunction. The 2D-CNN model demonstrated superior performance in terms of classification accuracy whencompared with the KNN approach. The 2D-CNN model attained a classification accuracy of 94%, whilethe KNN method achieved 88%. The findings indicate the efficacy of deep 2D-CNNs (Convolutional NeuralNetworks) in the classification of hyperspectral land cover. The results also demonstrate the networks’suitability for implementation in large-scale remote sensing projects.References
Bioucas-Dias, Jos M., Plaza, Antonio, Camps-
Valls, Gustavo, Scheunders, Paul, Nasrabadi,
Nasser, Chanussot, Jocelyn, Hyperspectral remote
sensing data analysis and future challenges, vol.
, no. 2, pp. 6{36, IEEE, 2013.
Chen, Yushi, Lin, Zhouhan, Zhao, Xing, Wang,
Gang, Gu, Yanfeng, Deep learning-based classi-
cation of hyperspectral data, vol. 7, no. 6, pp.
{2107, IEEE, 2014.
Ham, Jisoo, Chen, Yangchi, Crawford, Melba M.,
Ghosh, Joydeep, Investigation of the random for-est framework for classication of hyperspectral
data, vol. 43, no. 3, pp. 492{501, IEEE, 2005.
Ma, Li, Crawford, Melba M., Tian, Jinwen, Lo-
cal manifold learning-based k-nearest-neighbor for
hyperspectral image classication, vol. 48, no. 11,
pp. 4099{4109, IEEE, 2010.
Li, Jun, Bioucas-Dias, Jos M., Plaza, Antonio,
Spectral{spatial hyperspectral image segmentation
using subspace multinomial logistic regression and
Markov random elds, vol. 50, no. 3, pp. 809{823,
IEEE, 2011.
Melgani, Farid, Bruzzone, Lorenzo, Classication
of hyperspectral remote sensing images with sup-
port vector machines, vol. 42, no. 8, pp. 1778{
, IEEE, 2004.
Mughees, Atif, Tao, Linmi, Ecient deep auto-
encoder learning for the classication of hyper-
spectral images, in Proceedings of the 2016 Inter-
national Conference on Virtual Reality and Visu-
alization (ICVRV), pp. 44{51, IEEE, 2016.
Chen, Yushi, Zhao, Xing, Jia, Xiuping, Spectral{
spatial classication of hyperspectral data based on
deep belief network, vol. 8, no. 6, pp. 2381{2392,
IEEE, 2015.
Chen, Yushi, Jiang, Hanlu, Li, Chunyang, Jia, Xi-
uping, Ghamisi, Pedram, Deep feature extraction
and classication of hyperspectral images based on
convolutional neural networks, vol. 54, no. 10, pp.
{6251, IEEE, 2016.
Farooq, Adnan, Hu, Jiankun, Jia, Xiuping, Weed
classication in hyperspectral remote sensing im-
ages via deep convolutional neural network, in
Proceedings of IGARSS 2018 - IEEE Interna-
tional Geoscience and Remote Sensing Sympo-
sium, pp. 3816{3819, IEEE, 2018.
Gmez-Chova, Luis, Tuia, Devis, Moser, Gabriele,
Camps-Valls, Gustau, Multimodal classication
of remote sensing images: A review and future
directions, vol. 103, no. 9, pp. 1560{1584, IEEE,
Ahmad, Muhammad, Shabbir, Sidrah, Roy,
Swalpa Kumar, Hong, Danfeng, Wu, Xin, Yao,
Jing, Khan, Adil Mehmood, Mazzara, Manuel,
Distefano, Salvatore, Chanussot, Jocelyn, Hy-
perspectral image classicationTraditional to deep
models: A survey for future prospects, vol. 15, pp.
{999, IEEE, 2021.
Kong, Yi, Wang, Xuesong, Cheng, Yuhu,
Spectral{spatial feature extraction for HSI classi-
cation based on supervised hypergraph and sample
expanded CNN, vol. 11, no. 11, pp. 4128{4140,
IEEE, 2018.
Yuan, Yuan, Zhu, Guokang, Wang, Qi, Hyper-
spectral band selection by multitask sparsity pur-
suit, vol. 53, no. 2, pp. 631{644, IEEE, 2014.
Sharma, Vivek, Diba, Ali, Tuytelaars, Tinne,
Van Gool, Luc, Hyperspectral CNN for im-
age classication & band selection, with ap-
plication to face recognition, Technical report
KUL/ESAT/PSI/1604, KU Leuven, ESAT, Leu-
ven, Belgium, 2016.
Zhao, Wenzhi, Du, Shihong, Spectral{spatial fea-
ture extraction for hyperspectral image classica-
tion: A dimension reduction and deep learning
approach, vol. 54, no. 8, pp. 4544{4554, IEEE,
Yu, Shiqi, Jia, Sen, Xu, Chunyan, Convolutional
neural networks for hyperspectral image classi-
cation, vol. 219, pp. 88{98, Elsevier, 2017.
Vaddi, Radhesyam, Manoharan, Prabukumar,
Hyperspectral image classication using CNN
with spectral and spatial features integration, vol.
, pp. 103296, Elsevier, 2020.
Lee, Hyungtae, Kwon, Heesung, Going deeper
with contextual CNN for hyperspectral image clas-
sication, vol. 26, no. 10, pp. 4843{4855, IEEE,
Bera, Somenath, Shrivastava, Vimal K., Satap-
athy, Suresh Chandra, Advances in Hyperspec-
tral Image Classication Based on Convolutional
Neural Networks: A Review, vol. 133, no. 2,
CMES-Computer Modeling in Engineering & Sci-
ences, 2022.
Huang, Jiadong, He, Hongyuan, Lv, Rulin,
Zhang, Guangteng, Zhou, Zongxian, Wang, Xi-
aobin, Non-destructive detection and classica-
tion of textile bres based on hyperspectral imag-
ing and 1D-CNN, vol. 1224, pp. 340238, Elsevier,
Bird, Jordan J., Faria, Diego R., Manso, Luis J.,
Ayrosa, Pedro P. S., Ekart, Aniko, A study on
CNN image classication of EEG signals repre-
sented in 2D and 3D, vol. 18, no. 2, pp. 026005,
IOP Publishing, 2021.
Farokhah, Lia, Sarno, Riyanarto, Fatichah, Chas-
tine, Simplied 2D CNN architecture with channel
selection for emotion recognition using EEG spec-
trogram, vol. 11, pp. 46330{46343, IEEE, 2023.
Diakite, Alou, Jiangsheng, Gui, Xiaping, Fu,
Hyperspectral image classication using 3D 2D
CNN, vol. 15, no. 5, pp. 1083{1092, Wiley On-
line Library, 2021.
Luo, Yanan, Zou, Jie, Yao, Chengfei, Zhao, Xi-
aosong, Li, Tao, Bai, Gang, HSI-CNN: A novel
convolution neural network for hyperspectral im-
age, in Proceedings of the 2018 International Con-
ference on Audio, Language and Image Processing
(ICALIP), pp. 464{469, IEEE, 2018.
Feng, Fan, Wang, Shuangting, Wang, Chunyang,
Zhang, Jin, Learning deep hierarchical spatial{
spectral features for hyperspectral image classi-
cation based on residual 3D-2D CNN, vol. 19, no.
, pp. 5276, MDPI, 2019.
Zhang, Haitao, Meng, Lingguo, Wei, Xian, Tang,
Xiaoliang, Tang, Xuan, Wang, Xingping, Jin, Bo,
Yao, Wei, 1D-convolutional capsule network for
hyperspectral image classication, arXiv preprint
arXiv:1903.09834, 2019.
Nguyen, Duc Huy, Chao, Paul C.-P., Chung,
Chih-Chieh, Horng, Ray-Hua, Choubey, Bhaskar,
Detecting atrial brillation in real time based on
PPG via two CNNs for quality assessment and
detection, vol. 22, no. 24, pp. 24102{24111, IEEE
Sensors Journal, 2022.
DOI:
https://doi.org/10.31449/inf.v49i34.9480Downloads
Published
How to Cite
Issue
Section
License
I assign to Informatica, An International Journal of Computing and Informatics ("Journal") the copyright in the manuscript identified above and any additional material (figures, tables, illustrations, software or other information intended for publication) submitted as part of or as a supplement to the manuscript ("Paper") in all forms and media throughout the world, in all languages, for the full term of copyright, effective when and if the article is accepted for publication. This transfer includes the right to reproduce and/or to distribute the Paper to other journals or digital libraries in electronic and online forms and systems.
I understand that I retain the rights to use the pre-prints, off-prints, accepted manuscript and published journal Paper for personal use, scholarly purposes and internal institutional use.
In certain cases, I can ask for retaining the publishing rights of the Paper. The Journal can permit or deny the request for publishing rights, to which I fully agree.
I declare that the submitted Paper is original, has been written by the stated authors and has not been published elsewhere nor is currently being considered for publication by any other journal and will not be submitted for such review while under review by this Journal. The Paper contains no material that violates proprietary rights of any other person or entity. I have obtained written permission from copyright owners for any excerpts from copyrighted works that are included and have credited the sources in my article. I have informed the co-author(s) of the terms of this publishing agreement.
Copyright © Slovenian Society Informatika
