Open AccessAn easy method for barchan dunes automatic extraction from multispectral satellite data
Author(s) -
Ali Aydda,
Omar F. Althuwaynee,
Binod Pokharel
Publication year - 2020
Publication title -
iop conference series. earth and environmental science
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.179
H-Index - 26
eISSN - 1755-1307
pISSN - 1755-1315
DOI - 10.1088/1755-1315/419/1/012015
Subject(s) - multispectral pattern recognition , multispectral image , remote sensing , satellite , principal component analysis , extraction (chemistry) , computer science , geology , artificial intelligence , pattern recognition (psychology) , engineering , chemistry , chromatography , aerospace engineering
This work presents an easy method for barchan dunes automatic extraction from multispectral satellite data. The proposed method based on unsupervised classifications of commonly used bands for sand dunes mapping in literature. First, the collected data were atmospherically and spatially enhanced. Moreover, each selected band (band ratio or redness index or crust index) were filtered using low-pass (3x3) filter and transformed with original image (non-filtered) by using principal component analysis (PCA). Additionally, the classifications were achieved for each selected band by using three different algorithms (K-means, Expectation Maximization (EM), and IsoData) after data transformation. Eventually, the obtained maps were segmented and compared with natural colour image. The results indicate that unsupervised classification of crust index selected band, which achieved by IsoData algorithm, presents high performance for barchan dunes detection.