Klasifikasi Tutupan Lahan Menggunakan Convolutional Neural Network pada Citra Satelit Sentinel-2

Eka Miranda, Mediana Aryuni

Abstract


Abstrak


Perencanaan dan pemantauan penggunaan lahan masih menjadi masalah di banyak negara berkembang seperti Indonesia. Teknologi konvensional dan observasi lapangan yang digunakan saat ini memiliki kendala terkait dengan wilayah Indonesia yang luas. Deep learning dan citra satelit dapat digunakan untuk mendukung otomatisasi pemantauan dan klasifikasi tutupan lahan. Tujuan penelitian ini adalah mengembangkan metode klasifikasi berdasarkan Convolutional Neural Network (CNN) dan citra satelit Sentinel-2. Penelitian ini menggunakan: fitur spektral, indeks spektral, fitur spasial sebagai input dan definisi kelas tutupan lahan dari dokumen RSNI-1 Badan Standar Nasional Indonesia untuk Klasifikasi Kelas Penutupan Lahan dalam Penafsiran Citra Optis Resolusi Sedang sebagai dasar untuk ekstraksi fitur. Area citra penelitian terletak di daerah Semarang. Alur penelitian meliputi: Mendefinisikan kelas hutan berdasarkan dokumen RSNI-1, Mengekstraksi fitur optik citra berdasarkan definisi kelas tutupan lahan, Mengekstraksi fitur citra dari citra satelit Sentinel-2, dan Mengklasifikasikan fitur objek citra menggunakan pengklasifikasi CNN. Segmentasi citra menghasilkan 2072 objek dengan menggunakan eCognition. Label kelas data pengujian ditentukan menggunakan pelatihan semi terawasi. Klasifikasi tutupan lahan dilakukan menggunakan Convolutional Neural Network untuk mengklasifikasikan objek ke kelas tutupan lahan hutan kering primer, hutan kering sekunder, hutan tanaman, padang rumput, pemukiman, badan air atau tanah terbuka. Metode CNN yang digunakan mampu mengklasifikasi dengan baik yaitu overall accuracy 98.4% untuk data pelatihan dan data pengujian, serta rata-rata nilai akurasi user’s accuracy dan producer’s accuracy lebih besar dari 95% untuk tiap kelas tutupan lahan.


Kata kunci: Convolutional Neural Network (CNN), klasifikasi tutupan lahan, Sentinel-2,

Abstract


Land cover planning and monitoring are problems in many developing countries and Indonesia. Conventional technology and field observations have constraint related to the wide coverage area in Indonesia. Deep learning and satellite imagery can be used to support automation of monitoring and land cover classification and planning. The research aimed to study a classification method based on Convolutional Neural Network (CNN) and Sentinel-2 satellite imagery. Spectral indices and spatial features were used as an input whereas the RSNI-1 document Indonesian National Standards Agency for Land cover classification in Medium Resolution Optical Images Interpretation as a basis for feature extraction. The study area covered land cover in Semarang area. The research flow consists of Defining forest classes based on the RSNI-1 document, Extracting optical features of imagery from the definition of land cover class, Extracting image features from Sentinel-2 satellite imagery, and Classifying the image object feature using the CNN. Image segmentation produced 2072 objects using eCognition software. Training data label for each class determined using a semi-supervised training approach. The land cover classification performed using the Convolutional Neural Network to classify the object into land cover class namely primary dry forest, secondary dry forest, plantation forest, grassland, settlement, waterbody or bare land. This research showed the CNN method performs high accuracy value result: overall accuracy of 98.4% for both training data and testing data. Avarage accuracy value for User’s accuracy and producer’s accuracy are more than 95% for each land cover classes.


Keywords: Convolutional Neural Network (CNN), land cover classification, Sentinel-2


Full Text:

PDF

References


M. Kampffmeyer, A. Salberg, R. Jenssen, “Urban Land Cover Classification with Missing Data Using Deep Convolutional Neural Networks”, in IEEE International Geoscience and Remote Sensing Symposium, 2017, pp. 5161-5164.

Q. Weng, Z. Mao, J. Lin, X. Liao, “Land-Use Scene Classification Based on A CNN Using A Constrained Extreme Learning Machine”, International Journal of Remote Sensing, pp. 1-18. 2018.

L. Ma, X. Liu, Y. Zhang, G. Yin, B.A Johnson, “Deep Learning in Remote Sensing Applications: A Meta Analysis and Review” ISPRS Journal of Photogrammetry and Remote Sensing, vol. 152, pp. 166-177, 2019.

K. He, S. Zhang, S. Ren, J. Sun, “Deep Residual Learning for Image Recognition” in IEEE International Conference on Computer Vision and Pattern Recognition, 2016, pp. 770-778.

K. Simonyan, A. Zisserman, “Very Deep Convolutional Networks for Large-Scale Image Recognition”, in IEEE International Conference on Learning Representation, 2015, pp. 1-14.

R. Girshick, J. Donahue, T. Darrell, J. Malik, “Region-Based Convolutional Networks for Accurate Object Detection and Segmentation”, IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 38, issue. 1, pp. 142-158, 2016.

J. Redmon, S. Divvala, R. Girshick, A. Farhadi, “You only look once: Unified, Realtime Object Detection” in IEEE International Conference on Computer Vision and Pattern Recognition, 2016, pp. 779-788.

Y. LeCun, K. Kavukcuoglu, C. Farabet, C. “Convolutional Networks and Applications in Vision”, in IEEE International Symposium on Circuits and Systems, 2010, pp. 253–256.

K. Uddin, H. Gilani, M. Murthy, R. Kotru, F.M. Qamer, “Forest Condition Monitoring Using Very-Highresolution Satellite Imagery in A Remote Mountain Watershed in Nepal”, Mountain Research and Development, vol. 35, issue. 3, pp. 264–278, 2015.

W. Zhao, S. Du, W.J. Emery, “Object-Based Convolutional Neural Network for High-Resolution Imagery Classification”. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, vol. 10. issue. 7, pp. 3386–3396, 2017.

N. Kussul, M. Lavreniuk, S. Skakun, A. Shelestov, “Deep Learning Classification of Land Cover and Crop Types using Remote Sensing Data” IEEE Geoscience and Remote Sensing Letters, vol. 14, issue. 5, pp. 778–782, 2017.

Y. Zhu, S. Newsam, “Land Use Classification Using Convolutional Neural Networks Applied to Ground-Level Images, in the 23rd SIGSPATIAL International Conference on Advances in Geographic Information Systems, 2015, pp. 1-4.

Badan Standardisasi Nasional, “RSNI-1 Standar Nasional Indonesia. Kelas Penutupan Lahan Dalam Penafsiran Citra Optis Resolusi Sedang”, Badan Standardisasi Nasional, 2014. [Online] Available: http://appgis.menlhk.go.id/appgis/download.aspx?status=view&filename=RSNI_1b_KELAS%20PENUTUPAN%20LAHAN%20DALAM%20PENAFSIRAN%20CITRA%20OPTIS%20RESOLUSI%20SEDANG.pdf&fileFullName=E:%5Cwebgisapp%5CDownload%5CPeraturan%20Terkait%5CRSNI_1b_KELAS%20PENUTUPAN%20LAHAN%20DALAM%20PENAFSIRAN%20CITRA%20OPTIS%20RESOLUSI%20SEDANG.pdf

E. Mandanici, G. Bitelli, “Preliminary comparison of sentinel-2 and landsat 8 imagery for a combined use”, Remote Sensing vol. 8, issue.12, pp. 1014, 2016.

A Canada Centre for Remote Sensing, Fundamentals of Remote Sensing, A Canada Centre for Remote Sensing, 2016.

Sentinel-2 EO. “Sentinel Hub Data”, [Online]. Available: https://www.sentinel-hub.com/explore/data/.

H. Saadat, J. Adamowski, R. Bonnell, F. Sharifi, M. Namdar, S. Ale-Ebrahim,”Land use and land cover classification over a large area in Iran based on single date analysis of satellite imagery’, ISPRS Journal of Photogrammetry and Remote Sensing, vol. 66, issue. 5, pp. 608–619, 2011.

GISGeography, “Supervised and Unsupervised Classification in Remote Sensing”, GISGeography, 2018. [Online]. Available: https://gisgeography.com/supervised-unsupervised-classification-arcgis/

Trimble Geospatial, “eCognitionTransform Data to Information”. [Online]. Available: https://geospatial.trimble.com/products-and-solutions/ecognition.

Trimble Geospatial, Benefits of Object-based Feature Extraction. [Online]. Available: https://geospatial.trimble.com/blog/benefits-object-based-feature-extraction.

G.B. Scofield, E. Pantaleao, R.G. Negri, “A Comparison of Accuracy Measures for Remote Sensing Image Classification: Case Study in An Amazonian Region using Support Vector Machine”, International Journal of Image Processing, vol. 9, issue. 1, pp. 11–21, 2015.




DOI: https://doi.org/10.32520/stmsi.v10i2.1226

Article Metrics

Abstract view : 2091 times
PDF - 1125 times

Refbacks

  • There are currently no refbacks.


Creative Commons License
This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.
https://section.iaesonline.com/akun-pro-kamboja/https://journals.uol.edu.pk/sugar-rush/http://mysimpeg.gowakab.go.id/mysimpeg/aset/https://jurnal.jsa.ikippgriptk.ac.id/plugins/https://ppid.cimahikota.go.id/assets/demo/https://journals.zetech.ac.ke/scatter-hitam/https://silasa.sarolangunkab.go.id/swal/https://sipirus.sukabumikab.go.id/storage/uploads/-/sthai/https://sipirus.sukabumikab.go.id/storage/uploads/-/stoto/https://alwasilahlilhasanah.ac.id/starlight-princess-1000/https://www.remap.ugto.mx/pages/slot-luar-negeri-winrate-tertinggi/https://waper.serdangbedagaikab.go.id/storage/sgacor/https://waper.serdangbedagaikab.go.id/public/images/qrcode/slot-dana/https://siipbang.katingankab.go.id/storage_old/maxwin/https://waper.serdangbedagaikab.go.id/public/img/cover/10k/