Vision Localization Algorithm for Unmanned Aerial Vehicle Using Active Landmark

According to the problem of target recognition and localization for UAV, a target recognition and localization algorithm based on active landmark and mono vision is proposed. Two red LED lights are used in the active landmark. The flight control computer can transmit the control commands to the landmark from wireless for switch in many landmarks, the image is preprocessed with undistortion, median filtering and morphological filtering. Finally, the target is located by the contours of the landmark. The central point of the landmark in the camera coordinate can be reconstructed with the parameters of camera calibration and the coordinates in pixel plane. Then the relative location and yaw angle between UAV and target in the world coordinate system can be got. Data is optimized with the Kalman filter. The result of the experiment shows that the error of the algorism is little. This algorism has strong stability and accuracy.