Research on Horizon Detection Algorithm Based on Dark Channel Prior for Images Captured by UAV

Abstract

To detect the horizon for the UAV attitude measurement based on the visual navigation, a horizon detection algorithm based on the dark channel prior in foggy weather is proposed. This algorithm firstly uses the principle of dark channel prior for the haze removal to get the dark channel image, then uses horizontal and vertical Sobel operator to carry on edge detection on the dark channel image, getting the binary image. Finally, detect the horizon through the method of rotating image. Experiment result shows that, this algorithm can effectively detect the horizon and is robust to some complex scenes.