Algorithms for IR data

ABSTRACT: We and our team partner Emu Solutions, a University of Notre Dame spin-off, will implement and ultimately deliver ISiS_PFA (Image Stabilization in Silicon_PFA), an electronics hardware implementation of our recently-invented Polynomial Factor Analysis (PFA) jitter mitigation algorithm. This algorithm works, not on an OPIR payload"s beam stabilization front-end, but rather on the stream of collected sensor images to reconstruct the jitter-free image stream. We invented PFA to surmount the shortcomings of previous jitter mitigation algorithms in the face of demanding OPIR applications, including PCA, frame registration approaches, and simplistic consumer-grade electronic image stabilization. This proposed effort will thoroughly assess and demonstrate the performance and utility of the PFA algorithm for OPIR and HyperTemporal Imaging application. From the get-go, we will endeavor to translate the PFA algorithm into a real-time hardware implementation (GPU and/or FPGA) suitable for insertion into SMCs Falcon-Shield ground-based processing center (previously the Aerospace Fusion Center) at Schriever AFB, and ultimately into real-time spacecraft data-processing firmware. Emu provides a unique"Processing-in-Memory"(PIM) inspired approach that in effect provides a massively multithreaded mobile-threadlet"operating system"on an FPGA. This is key for parallelizing the non-streamable PFA algorithm kernels while maintaining the requisite memory-access throughput. BENEFIT: Our proposed product, ISiS_PFA Image Stabilization in Silicon based on our PFA algorithm, offers cost and performance benefit for various OPIR applications and across DoD imaging surveillance and targeting applications.