Abstract:

In micro air vehicle’s visual navigation, splitting the image into two parts (sky and ground) is a fundamental problem. This is the prerequisite of the visual navigation system’s closedloop control. However, the foundation of sky-ground segmentation is the extraction of the characteristic line. By using the design philosophies of bottom-up and top-down respectively, both generalized Hough transform method and multiscale least square based line extraction method are realized in this paper.

Key words: Bayesian , classifier,  , aircraft , visual , navigation,  , generalized , Hough , transform