Research On Vision-based Autonomous Landing Of Unmanned Helicopter

Unmanned Helicopter is being a research focus,because of low requirement of the rising and landing conditions,well stability,well controllability and carrying out complex tasks ability and so on.Navigation system is the core of Unmanned Helicopter,Vision-based navigation system which becomes the preferred way because of providing more accurate position information.This paper primary research camera model and calibration,the posture parameters,landing object real-time tracking of Unmanned Helicopter based on visual navigation system.The main idea of this paper takes advantage of the recognition and tracking of Unmanned Helicopter System and completes the process of autonomous landing of Unmanned Helicopter,based on visual navigation system.Its content is:(1)The landing object’s pretreatment.By collecting images in imaging equipment,process the collected images’ histogram graying,noise filtering,edge detection.At the same time,based on the edge detection operator’s lack in filtering out false edge,design a improving Canny edge operator to determine two thresholds,where the grayscale image is divided into three regions,calculate the maximum variance’s size to determine the best double threshold by calculating the three areas’ average gray and regional gray probability.Through the experimental simulation,the dissertation proved that the method can effectively enhance the edge of the reinforcement and remove the edge of the pseudo.(2)According to the space geometry relation in 2d images and 3d scene,define corresponding coordinate system and set up camera imaging model.According to the camera imaging model,perform camera calibration and get the internal parameters of camera equipment.(3)After image pretreatment,mark and extraction corner by the Harris corner points method,which input the position equation of Unmanned Helicopter to solve the attitude and position of Unmanned Helicopter.Simulation proving: The posture of Unmanned Helicopter is correct and accurate.(4)By the image contour extraction technology,realize object detection and recognition based on visual system.On the basis of related references,design acolorless kalman filter based on color noise to object real-time tracking.Through the simulation,verify the feasibility of object tracking and identification of Unmanned Helicopter.The results show that the method is accurate and feasible.