A Robust Vision-Based Skyline Detection Algorithm Under Different Weather Conditions

Abstract

The skyline is a piece of important reference information in the automatic flight control system of an unmanned vehicle. Currently, most skyline detection algorithms assume that the skyline has distinct edge features that can be located in the image by edge detection. This method of detection, however, is only applicable when the skyline is obvious. Its detection result in images with an indistinct skyline is affected by the threshold value, such as on cloudy or heavily foggy days, when clear skyline features cannot be found. To find both distinct and indistinct skylines in the images, this paper proposes a vision-based skyline detection algorithm, in which the skyline is located by analyzing image brightness variations. This method is able to identify nonlinear skyline profiles in scenes with a clear skyline, and find the interface region between the sky and earth in scenes with an indistinct skyline, to estimate its location. According to the experimental results, the algorithm correctly finds the skyline profile in 97% of the test images, making it ideal to provide skyline reference information for the automatic flight control systems of unmanned vehicles.