无人机视觉导航中的地平线检测算法研究

Visual navigation is a new generation of navigation technology to obtain carrier navigation parameters through image from visual sensor, and it has the advantages of high precision, high sensitivity etc. UAV visual navigation technology development have important meanings for UAV control technology and modern information war. This paper attempts to apply computer vision to UAV autonomous navigation in the landing phase and to comprehensively utilize a variety of image processing techniques based on images acquired during landing phase in research on horizon detection algorithm in UAV visual navigation. The main works include preprocessing, edge detection, the initial horizon extraction by line detection based on the Hough transform and horizon extraction using a fitting algorithm. The experimental results show that the proposed horizon detection procedure can extract the horizon completely and correctly, the UAV attitude detection can be completed based on such procedure, which may assist for the mission of visual navigation.