Hyperspectral Imaging Simulator and Applications for Unmanned Aerial Vehicles

Thumbnail Image
Atarita, Fadhli
Hyperspectral Imaging , UAVs , Simulator , Remote Sensing
Mineral exploration utilizing hyperspectral imaging sensors aboard low-altitude airborne platforms is still a relatively new and developing research topic. The main reason is the difficulty of finding the optimal combination of platform, sensor, and survey design for specific targets in various geological settings. This research firstly aims to examine the reliability of an industry-standard workflow for hyperspectral imaging using an unmanned aerial vehicle (UAV) for a mineral exploration target in Cuprite, Nevada. A versatile hyperspectral imaging (HSI) simulator called HYSIMU has been developed that can be used to simulate different scenarios based on customizable flight and target parameters. Random mineral reflectance spectra are assigned to synthetic digital elevation models (DEMs) and several flight scenarios are simulated on those DEMs using different sensors and selectable survey parameters such as altitude, speed, and sun position with options to add spatial noise, spectral noise, and terrain shadow effects. The synthetic hyperspectral data generated from each scenario are further processed using the workflow established earlier in this research to determine the target minerals and to create classification maps. In this study, three ground truth scenarios with different levels of complexity were simulated with varying survey parameters such as altitude (10 to 200 meters), velocity (1 to 100 m/s), and sun elevation (0°, 45°, and 90°). A sensitivity study is performed using four metrics to compare the ground truth data and the obtained classification maps. The results showed that HYSIMU has great potential as a tool to optimize hyperspectral imaging in mineral explorations by finding the right combinations of flight parameters, primarily the altitude and flight speed which affect the survey time and total cost. There are opportunities for further development of HYSIMU to make it more versatile. It was also implemented to test the detectability of agricultural targets in different soil conditions. Lastly, the use of satellite-based Land Surface Temperature (LST) as an earthquake precursor is examined and its potential to be combined with hyperspectral imaging is discussed.
External DOI