Our institute member Assoc. Prof. Gülsen Taskin with her master student, Enes Oguzhan Alatas, published their study titled “An earthquake damage identification approach from VHR image using mathematical morphology and machine learning” in Neural Computing and Applications.
To access this article:
The purpose of this study is to investigate the use of morphological profiles and attribute profiles as additional spatial features for very high resolution post-event satellite images in order to improve the accuracy of earthquake-induced damage assessment. The efficacy of these features is evaluated by applying a feature selection algorithm and testing the resulting feature subset using different classifiers. The study aims to determine the potential of these profiles to enhance the quality of the thematic map, but also to consider the potential limitations of the classification model in terms of generalization for larger areas.