Research On Detection Identification Algorithm Of Drogue Images And Flight Demonstration

Air refueling, known as the "air force multiplier", is an effective means of addressing the short flight of aircraft, and it has always been attached great importance by the aviation developed countries. At present, Chinese air refueling always relys on the receiver aircraft pilots to perceive the relative position between the oil plug and drogue docking which is in front of plug. The operation is difficult and dangerous. Thus it is necessary to develop a set of assistant aerial refueling guidance system.This thesis studies the detection, recognition algorithm of drogue target using image technology; and does further study of flight demonstration based on small rotorcraft.Firstly, the detection algorithm of the aerial refueling's drogue target is studied. After the preprocessing operations of image calibration correction, thresholding and filtering, this paper presents a refueling drogue detection method based on Hough-LS, which locates the precise positioning of the cone sleeveSecondly, the pattern recognition technology of drogue target is studied. The thesis makes use of Gabor filter to extract drogue target feature, and does dimensionality reduction based on principal component analysis. On this basis, drogue target recognition has been put forword with the use of Fisher linear classifier. After the analysis, experimental results show the effectiveness of the algorithm.Futhermore, the design of the flight demonstration system which is based on small rotorcraft is proposed. The thesis gives hardware platform design requirements which is the carrier of flight demonstration system, and analyzes its hardware architecture. Considering with actual situation of the aerial refueling, key schemes are proposed, such as whole system, the communication link and visual navigation segment flight stage.Finally, the thesis builds flight demonstration experiment platform, and launchs a number of experimental tests, such as flying target detection under dynamic conditions, the four-rotor aircraft autopilot docking and docking accuracy. Experiment results show that the algorithm is correct under conditions of dynamic flying, with better detection stability.