Real-time Computational Enhancement of Video Streams

Videos captured from aerial, satellite, maritime, and ground platforms are often degraded by the effects of atmospheric disturbances, detector noise, camera jitter, compression artifacts, and imperfect digital post-processing. Nanohmics’ existing Real-time Online GPU Enabled (ROGUE) software is designed to use fast, parallel multiframe processing algorithms to compensate for temporally-varying image defects including atmospheric turbulence, motion blur, compression artifacts, and general noise. It was originally developed to mitigate the effects of isoplanatic turbulence in video streams of ground-based telescopes on low orbit space objects, and later extended to mitigate some of the image defects and scene types in typical air-to-ground imagery. Nanohmics, in partnership with Johns Hopkins University/Applied Physics Laboratory, proposes to develop and validate an advanced version of ROGUE to improve compatibility with general distressed video from maritime and airborne collectors and reduce processing latencies. The proposed extensions are designed to relax assumptions about atmospheric and motion blurs, add support for color imagery, improve the quality of enhancements for IR imagery, automate the selection of optimal configuration parameters for general scenes, and increase the processing speeds. The capabilities of software prototypes will be demonstrated on real and synthetically-degraded maritime and airborne imagery already available to the team.