Novel Accelerator Architectures for HSI Detection and Identification

Utilization of specialized COTS processors for high-throughput functions of hyperspectral detection processing will enable the next generation of small, lightweight UAVs to provide target detection and identification currently only available on larger and more expensive aircraft. Next generation UAV platforms will require sensor systems with smaller footprints and lighter weight than the current generation of Hyperspectral sensor systems. Space Computer Corporation (SCC) proposes to enhance the computational performance of key HSI algorithms by utilizing the capabilities of Graphics Processor Units (GPUs). This proposal describes an approach that couples the power of GPUs with simple PC-based processing architectures to significantly reduce the size of airborne, real-time processors. This use of GPUs, coupled with the recent development of small HSI instruments, will enable cost-effective use of small, inexpensive platforms to support target detection and identification applications previously reserved for large UAV or airborne platforms.