Land Use Land Cover Mapping From Sentinel-1, Sentinel-2 and Fused Sentinel Images Based on Machine Learning Algorithms

Authors: ,

Abstract: Mapping land use/land cover is a challenge given the diversity of sensors available. This paper aims to use and evaluate the contribution of multisensors classification based on machine learning classifiers (neural network and support vector machine) since traditional approaches always suffer from the hand-designed features and misclassification of boundary pixels. pre-processing of the SAR data(Sentinel 1-A) has been performed. Data have been calibrated, terrain corrected, and filtered by a 5x5 kernel using gamma map approach. Sentinel-2 has been coregistered to Sentinel-1-A. Then principal component analysis PCA fusion was used to fuse multispectral and radar images. Support Vector Machines (SVM) method and neural network have been implemented as a supervised pixel based image classification to classify Sentinel-1 , Sentinel-2 and fused sentinel images During the classification, different scenarios(seven) have been applied to find out the performance of Sentinel-1 data. Different combinations of VV and VH polarizations have been analysed and the resulting classified images have been assessed using overall classification accuracy and Kappa coefficient. Results demonstrate that, combining opportunely dual polarization data, the overall accuracy increases up to 93.02% against 73.38% and 70.73% of using individual polarization VV and VH, respectively. Using more than one variable significantly increased the classification accuracy. Especially scenarios5, 6 and 7 has higher accuracies than other scenarios except scenario 4 which is the best. It was found that adding different types of data(attributes) to the source image gives a classifier more information to consider when assigning pixels to classes and improving classification accuracy.

International Journal of Computational and Applied Mathematics & Computer Science, E-ISSN: 2769-2477, Volume 1, 2021, Art. #3

PDF
XML
Cite
×
Citation Tools
WSEAS AMA
Lamyaa G. Taha, Rania E. Ibrahim, "Land Use Land Cover Mapping From Sentinel-1, Sentinel-2 and Fused Sentinel Images Based on Machine Learning Algorithms," International Journal of Computational and Applied Mathematics & Computer Science, vol. 1, pp. 12-23, 2021, DOI:
Lamyaa G. Taha, Rania E. Ibrahim. Land Use Land Cover Mapping From Sentinel-1, Sentinel-2 and Fused Sentinel Images Based on Machine Learning Algorithms. International Journal of Computational and Applied Mathematics & Computer Science. 2021;1:12-23.
Copy to Clipboard
Citation copied to Clipboard

Certification

---