Hyperspectral Image Processing Pipelines on Multiple Platforms for Coordinated Oceanographic Observation

摘要

The large size of hyperspectral images and the computational demands required for interpreting them complicates their acquisition, distribution, and utilization by remote agents. The Norwegian University of Science and Technology is building an ocean observation system, including the HYPSO-1 hyperspectral imaging satellite, to monitor ocean color in coastal waters. The images collected by HYPSO-1 will be used to inform an ensemble of robotic agents which locally monitor ocean conditions. Several agents will be unmanned aerial vehicles that likewise carry hyperspectral imagers. Hyperspectral images will be processed on-board the different agents, as well as on the ground. Here, we discuss the architecture, development, current status, and future opportunities of hyperspectral image processing pipelines on these platforms.