Image Processing and Machine Learning Methods for Analysis of Post Catastrophic Crisis Areas


Natural disasters crisis response is a burning topic for the remote sensing community, especially with the continuous increase of natural disaster occurrences in past decades. Opportune reaction times and high-quality crisis coverage data of such events can make the difference between a well-orchestrated crisis response and a disaster. In this regard earth observation data analysis can help decision making determine optimal strategies to apply crisis response where it is needed most. My goal is to investigate, evaluate and determine possible image processing and deep learning methods that help crisis response teams in the field and can be deployed directly during crisis response missions. The work is focused on multispectral and hyperspectral satellite imagery as base data, since image analysis methods developed, or models trained on this data can be extended to work with in-situ remote sensing methods such as drone imagery.