An Improved Algorithm for Horizon Detection Based on OSTU

Vision-based Navigation is an efficient method to estimate the UAV's attitude compared with traditional navigation for its unique advantages. The key step of vision-based navigation is to get useful information such as the horizon line from the image that taken from the camera installed in the UAV. In this paper, we introduce the attitude estimation method using the horizon from the image. We then propose an efficient algorithm based on OSTU to segment the image and use Hough transformation to detect the horizon line, so that we can get more accurate estimation for the UAV's attitude. The image processing results illustrate that the algorithm we proposed is efficient.