Image Deblurring in Roll Angle Estimation for Vision Enhanced AAV Control

In this paper a vision based sensing system is used to determine the importance of blur in images with regard to the estimation of planes angles. In particular the roll angle is estimated based on the perceived horizon line using gradient image algorithms. The gain in precision due to deblurring is estimated comparing trajectories obtained from on-board sensors with the vision based ones. Images from real experiments using a remote controlled plane are used.