@article{ref1,
title="Tropical cyclones intensity estimation by feature fusion and random forest classifier using satellite images",
journal="Journal of the Indian Society of Remote Sensing",
year="2022",
author="Kar, Chinmoy and Banerjee, Sreeparna",
volume="50",
number="4",
pages="689-700",
abstract="Tropical cyclones (TCs) are natural threats that cause huge damage in coastal regions and adjacent areas. Prior indication of TC intensity can prevent this loss. Hence, the intensity estimation of TC became an important task to reduce the risk. This paper proposes features fusion and classification techniques using infrared satellite images for intensity estimation. Features of TC images are extracted from Gray Level Cooccurrence Matrix, Muli-Block Local Binary Pattern, and Geometric features. These feature extraction techniques are applied on labeled TC images of the North Indian Ocean to generate features vector. The size of the feature vector is reduced to 93% using a correlation-based feature subset selection mechanism. Further, these reduced feature vectors are fused to generate a fused feature vector by the information gain-based feature selection method and Min-Max normalization. The Random Forest classifier is used on this final fused feature vector to classify TC images above 93% accuracy.<p /> <p>Language: en</p>",
language="en",
issn="0255-660X",
doi="10.1007/s12524-021-01477-5",
url="http://dx.doi.org/10.1007/s12524-021-01477-5"
}