Vector bank based target tracking via vision sensors in aviation systems

In this paper, we present a technique of detection and tracking of one or possibly more moving objects by implementing a "vector bank" into an H.264 based system-on-a-chip design architecture as a vision sensor in aviation systems, such as unmanned aerial vehicles (UAV). Most of the existing target-tracking algorithms are based on software solutions. The introduction of H.264 encoding chips provides a far better hardware solution by developing motion estimation blocks to detect macro-block movement up to 10 reference frames in streaming imaging data. By adding a vector bank with the proposed framework, a high-resolution H.264 based vision sensor could be easily developed and implemented for target tracking purpose in a real-time manner in aviation systems.