Application of Image Analysis and Machine Learning to Assist the Interpretation of Remotely Sensed Geospatial Data

This thesis presents a study that focuses on practical applications of image analysis and machine learning with a specific aim to derive usable outcomes for end-users. The proposed approach is based on integrating interactive visualisation with automated analysis to support data interpretation. This thesis presents three applications: (1) to map 3D orientations of geological structures from high resolution rock surface images captured from an unmanned aerial vehicle; (2) to assist the interactive mapping of lithological boundaries using rock surface images; and (3) to support the monitoring of vegetation recovery for rehabilitated mines using Landsat remote sensing images.