Automatic Target Detection and Tracking (ATD&T) Algorithms for Small Autonomous Projectiles

Given smooth motion of a projectile with an onboard imaging device, the displacement between focal plane centroid and desired aimpoint can be estimated via motion models customarily employed in video compres-sion. This supports target detection andmeasurement within a small predicted area of each image frame. Sensor output compression using an MPEG hardware codec would obviate potentially costly computation of a motion model, because motion vectors could be obtained from the MPEG data. Thisapproach, expected to yield significant computational sav-ings, would be based on the research team's highly successful research in still and video image compression. We have developed sophisticated, highly efficient algorithms for fast, accurate targetdetection. This requires no special hardware, and can be interposed in channels trans-mitting compressed or uncompressed imagery.Phase I will design and develop a joint hardware/software solution to maximally exploit imagery acquired from spinning projectiles, while minimizing power/space requirements. We propose to enhance and adapt our existing algorithms and software for earlyprototype demonstra-tion, and performance analysis of the proposed hardware/software given GFI models of the projectile imaging system and ballistic response. Phase-II will develop: (1) advanced motion-directed ATR software, to run on (2) low volume, lowpower hardware (COTS components) that sup-port real-time computation. Techniques developed under this topic are applicable to border protection, military target detection and image or video compression as well as domestic applications such as telecommunications. The commercial benefits for telecommunications areconsiderable. Other uses include faster detection of moving targets in law enforcement, and security applications.