Multi-Sensor Spectral Imaging of Geological Samples: A Data Fusion Approach using Spatio-Spectral Feature Extraction

Spectral imaging or hyperspectral reflectance mapping for mineral exploration sample analysis has evolved rapidly in the recent decade. A wide range of deployable sensors is available nowadays, providing high flexibility in spectral as well as in spatial resolution and coverage. However, the fusion of data from different customized setups and sensors is challenging and usually not conducted. In the following study, the integration of such multi-sensor datasets is demonstrated on data acquired from five commercially available hyperspectral sensors and a pair of RGB cameras. We present a workflow for the integrated image analyses using advanced machine learning methods and evaluate the procedure on a representative set of geological samples. Detailed mineralogical and spectral validation affirms the approach. The suggested workflow provides a new way for the integration of multi-source data, e.g., it allows a straight-forward integration of visible/near-infrared (VNIR), short-wave infrared (SWIR) and long-wave infrared (LWIR) data for sensors with highly different spatial and spectral resolution. Finally, we evaluate the benefits of different multi-sensor combinations for potential applications in mineral exploration.