Unmanned Aerial Vehicle (UAV) marker detection on a single-board computer

Abstract

In this project, a software was developed to detect UAV (Unmanned Aerial Vehicle) markers in real-time. There were some equipment used in the project, which were Ubuntu system, Robot Operating system, OpenCV, JN-mini5728 single board computer and Turtlebot2. The introduction of these equipment were given in the following part of this report. It also described in detail about several important steps to implement the vision tracking system. Firstly, camera calibration was a very important process, which determined the accuracy of the target. Then, the monocular camera used the HSV colour space to extract the marker on the target object, and then the marker detection program performed the GaussionBlur, erosion and dilation operations on the images, thereby obtaining a more accurate image of the target object. Next, the 3D pose estimation was used to analyse the trajectory of flight of target object, thus achieving target tracking. In the end of this report, I also gave some conclusion and suggestion on the future work.