Principal Components Transformation for Hyperspectral Data Compression

Period of Performance: 04/15/1998 - 10/15/1998

$100K

Phase 1 SBIR

Recipient Firm

Opto-knowledge Systems, Inc.
19805 Hamilton Ave Array
Torrance, CA 90502
Principal Investigator

Abstract

The principal components transformation will be applied to obtain real-time lossless compression of hyperspectral images, with compression factors of the order of 20-50. A large compression factor will permit UAVs and space borne sensors to gather and store far more data than before. This will not only reduce the operational costs of acquiring hyperspectral imagery drastically but also increasing the strategic significance of hyperspectral imaging technology. The computational requierements of the algorithm are of the order of several billion operations per second for data rates of 100MBits/s. Since no single current available processor is capable of handling the problem in real-time, dedicated hardware will be built using commercial off the shelf technology for real-time implementation. The architecture will exploit algorithm's inherent parallelism and combine an array of digital signal processors (DSPs) and field programmable gate arrays (FPGAs) to perform the task. The multiprocessor platform will be reconfigurable and can therefore be used to implement the entire class of parallel algorithms that fall within the single instruction multiple data (SIMD) paradigm.