RFCode: Robust Foveated Encoding

The Robust Foveated Encoding (RFCode) Phase 1 effort will improve battlefield capabilities and situational awareness by generating a detailed design and performance analysis for applying three emerging technologies to an advanced video compression system for unreliable tactical networks. RFCode uses foveated (variable-resolution) video coding to reduce data volume while maintaining operational effectiveness, enabling a superior compression ratio. It applies and extends distributed coding schemes inspired by new compressive sensing approaches and progressive coding techniques, providing robustness and graceful degradation under packet loss. Finally, RFCode algorithms are evaluated and optimized using novel video quality metrics that better measure subjective human video quality perception (and therefore operational effectiveness) than traditional error metrics. Performance is quantified with a combination of theoretic and empirical analyses, and compared against a baseline motion-JPEG (MJPEG) codec. By starting from firm theoretical foundations in foveation, compressive sensing, and video quality assessment, and by leveraging our team’s directly relevant past work in foveated image exploitation, 21st Century will address the Army’s operational need to reliably transmit imagery over low bandwidth, low reliability tactical wireless networks. This will increase the battlefield utility of tactical systems that transmit and exploit real-time battlefield imagery.