This thesis proposes the development of a machine learning-based framework to automate the cleaning and enhancement of bathymetry data. The research will explore supervised and unsupervised learning techniques to detect and remove anomalies, interpolate missing values, and improve overall data fidelity.
The goal is to reduce the manual workload while improving consistency and scalability across large datasets. The final solution could be integrated into existing hydrographic processing pipelines, offering a valuable tool for marine scientists and surveyors.
Point of contact: Arne Henrik Bechensteen, Kongsberg Discovery. arne.henrik.bechensteen@kd.kongsberg.com