Integrated learning-based and regularization-based super resolution for extreme MWIR image enhancement

OKSI and Northwestern University propose to develop a single-image super-resolution (SR) methodology for mid-wave infrared (MWIR) imagery that combines learning-based and regularization-based approaches to produce extreme enhancement of low-resolution images. We will also develop a detector-limited imaging system specifically designed to be used with the SR methodology for which even higher levels of image enhancement can be attained. The newly developed SR methodology will directly address the Navy’s desire for variable resolution MWIR imagery achieved through post-processing, as well as the imaging needs of other government agencies. During Phase-I, OKSI and Northwestern University demonstrated a proof-of-concept for the SR methodology and quantified the image enhancement that is attained. In Phase-II, we will build and demonstrate a prototype imaging system with near real-time post-processing for and ROI within the image.