Theory and Application of Foveated Acquisition and Tracking (TAFAT)

Foveated imagery is ubiquitous in nature, and foveating sensors offer the potential for superior performance in artificial vision systems. However, technical and user acceptance barriers caused in part by a lack of mature theory currently hinder the exploitation of foveated imagery on the battlefield. Our TAFAT Phase 1 experiments conclusively demonstrated that under reasonable assumptions and with advances in theory, foveated systems can consistently outperform uniform-resolution systems in multi-target acquisition and tracking. TAFAT Phase 2 builds on these successes to mature the theory and algorithms that let foveated sensors outperform uniform-resolution sensors. We will mature our methodology for quantifying the relative performance of foveated and uniform-resolution systems, and we will deliver a reusable evaluation framework with which AFRL can evaluate foveated exploitation approaches developed long after TAFAT. We will use our new theoretical developments to develop robust and flexible foveated acquisition, tracking, and recognition algorithms. We emphasize adaptive foveation techniques that automatically tune foveation characteristics to optimize acquisition and tracking, given current target and background characteristics. TAFAT enables the development of high-performance foveated targeting systems by jointly addressing theoretic and practical barriers. TAFAT leverages 21st Century’s specific experience and the internationally recognized expertise of our consultant, Professor Alan C. Bovik.