Dimension Reduction of Hyperspectral Data on Reconfigurable Computers

The objective of this work is to demonstrate the use of reconfigurable computing for on-board automatic processing of remote sensing data. The Field Programmable Processor Array (FPPA), a radiation tolerant alternative to Field Programmable Gate Arrays, developed at NASA/Goddard under ESTO funding, is the computation engine of our study, while preliminary feasibility studies are also performed on an hybrid reconfigurable architecture, the SRC-6. For this feasibility study, one of the first basic methodologies that is considered deals with dimension reduction of hyperspectral data. This processing represents one of the few most important first steps to be performed onboard, since it enables to reduce communication bandwidth, and make subsequent computations simpler, faster and more accurate. An adapted software environment developed for the FPPA is utilized for the design of this algorithm.