Underwater Sensing using Hyperspectral Imaging
Over the past decade, underwater hyperspectral imaging (UHI) has emerged as a promising seafloor mapping and monitoring technique. Underwater hyperspectral imagers may be deployed on a variety of sensor-carrying platforms (e.g. remotely operated vehicles and unmanned surface vehicles), which makes the technology applicable within fields such as marine biology, aquaculture, archaeology, geology and inspection of subsea structures. The UHI sensors we currently have at our disposal are push-broom scanners that cover the spectral range of 380-750 nm at a spectral resolution of <1 nm. During a push-broom acquisition of UHI data, the UHI sensor captures 1920 across-track pixels per exposure. Each of these pixels contains a contiguous light spectrum, which, as compared to a regular RGB (red, green, blue) value, provides better means of identifying and mapping objects of interest based on color information. As an example, many marine organisms spectrally differ more than what an RGB image would suggest, which means that higher spectral resolution is necessary for mapping their distribution and abundance accurately. The current presentation will both focus on past UHI results and some planned experiments. This includes biological, archaeological and geological applications of the technique, as well as an introduction to various hyperspectral data processing steps, such as reflectance conversion and georeferencing. We will also present a UHI platform that was built from two remotely operated vehicles and tested during a cruise in the Arctic in January.