The development of a high performance, high volume distributed hyperspectral processor and display system

While further research and development is still required in the field of hyperspectral processing, the state of the art is such that the user community can now obtain extremely useful information from the data. However, current systems are unable to meet the demand of operational users, and lag far behind hyperspectral sensors in their ability to process the large data volumes. The computational requirements for radiometric, geometric, and atmospheric adjustments combined with this large data volume means information extraction is slow. This imposes a limitation on the practical and operational use of the data. This situation will become more critical as data from a new generation of spaceborne hyperspectral sensors becomes available. There is a demonstrated need for a commercial, high throughput, hyperspectral processor. This paper describes aspects of an operational hyperspectral processor and display system required to meet the challenges of processing, analyzing and visualizing large volumes of data in a timely manner. Particular emphasis has been given to meeting the needs of the mining and exploration communities. The processor being developed will be easily expandable and will operate within an inexpensive distributed processing environment using highly optimized software designed to partition both the data and problem domains. The fact that hyperspectral processing is primarily pixel-based suggests that this approach can be highly effective in addressing the throughput required for an operational system.