Revised Version of a JCIT Paper-Comparison of Feature Point Extraction Algorithms for Vision Based Autonomous Aerial Refueling

Autonomous aerial refueling (AAR) is the most effective way to extend unmanned aerial vehicles’ (UAV) endurance and improve their mission performance. A monocular vision based relative pose estimation (MVRP) method is proposed to solve the problem of close-range and high-accuracy relative pose estimation, which is very important and critical for AAR. MVRP is a feature point tracking based method, which makes feature point extraction algorithm become a key issue for MVRP method. An MVRP virtual reality simulation system is set up based on Vega Prime. This paper carries out several simulation experiments to compare three feature point extraction algorithms within the MVRP method. The simulation results show that FAST algorithm is able to get a good balance between speed and accuracy, compared with the other two algorithms.