Vision-based relative altitude estimation of small unmanned aerial vehicles in target localization

This paper presents a UAV vision-based relative altitude estimation method using a given size (length) of the ground vehicle. In conjunction with a well known target localization technique, the proposed method relaxes the flat ground assumption and provides an independent approach of altitude correction. It offers possible improvement or expansion in small UAV applications in target localization, tracking, and terrain exploration missions, such as ground terrain mapping. In addition, the proposed kinematic method is reliable and computationally efficient. Its feasibility and effectiveness is further demonstrated by extensive experiments, including both stationary and moving target estimations.