Hyperspectral remote sensing and navigation in coastal environments using UAVs

Abstract

Imaging spectroscopy, or hyperspectral imaging, is a remote sensing technology that allows a user to gather information about an observed object based on its spectral reflectance. Traditionally, this technology was reserved for users with large financial resources. However, the commercialization of semiconductor image sensors based on Charge-Coupled Device (CCD) sensors has led to various companies producing commercially available imaging spectrometers [66, 127, 129]. Airborne imaging spectroscopy is an upcoming field that could potentially transform many industries. However, the high cost of commercially available imaging spectrometers remains a significant barrier, preventing widespread adoption and limiting the technology’s benefits to a smaller, more affluent user group. This thesis aims to show that an airborne imaging spectroscopy payload, resp. hyperspectral imaging payload can be built with Commercial Of-The-Shelf (COTS) components and be flown on small inexpensive Unmanned Aerial Vehicles (UAVs). It furthermore elaborates on the steps necessary to post-process and analyze the imaging spectroscopy data. While this work presented in this thesis focuses on Ocean Color (OC) mapping, which is accomplished by using imaging spectroscopy to monitor and analyze the spectral characteristics of ocean water, the developed payload can be adapted for a wide range of other applications. The work deals not only with the design and development of payloads for imaging spectroscopy. Operational considerations that can impact data quality and the calibration of the push-broom imaging spectrometer are also addressed. Special focus is given to georeferencing of hyperspectral- and Red-Green-Blue (RGB) data. This is achieved with a state estimation and georeferencing algorithm, which makes it possible to obtain directly georeferenced data from the rigidly mounted hyperspectral imager. Furthermore, data preparation and the associated post-processing pipeline as well as hyperspectral data analysis are presented. No commercial software was used specifically for the hyperspectral application for these purposes either. Additionally, the hyperspectral data product is compared with hyperspectral data from other agents. Another topic covered in this thesis is robust navigation in Global Navigation Satellite System (GNSS) denied areas. GNSS denied navigation is a topic that is becoming increasingly important. Situations, where no GNSS reception exists, can arise in various intentional or unintentional scenarios. Examples are deliberate jamming, unintended interference of the GNSS signal, or flying low in inaccessible, shielding terrain at high latitudes. Due to the increasing number of electronic devices that could interfere with the GNSS signal and due to the increasingly difficult security situation in Europe and other parts of the world, GNSS-denied navigation solutions become more important. This thesis presents a more robust version of an existing GNSS denied navigation solution.