Forward-looking Infrared Image Matching Navigation And Positioning Method

Inertial navigation/scene matching integrated navigation is an autonomous navigation system that does not rely on satellite navigation(GPS,Beidou),which has the characteristics of strong anti-interference ability and high positioning accuracy.In order to effectively improve the efficiency of imaging systems and enhance the ability of missiles to strike targets,the new generation of imaging seeker often adopts a forward-looking imaging mode.The forward-looking imaging has both target detection and navigation positioning functions.However,for matching and positioning,the forward-looking imaging will cause geometric distortion between the real-time image and the reference image,so the design of the matching algorithm and the positioning algorithm is more difficult.The thesis has deeply studied matching algorithm,positioning algorithm and combined filtering algorithm for practical application requirements,and obtained good experimental results.The main work of this thesis is as follows:A fusion heterogeneous image matching algorithm based on SIFT and LDB is proposed to improve the adaptability and matching accuracy of matching algorithm.Firstly,geometric correction and image enhancement preprocessing are performed on the original image.Then,bilateral filtering is used instead of the traditional Gaussian filter to construct the scale space pyramid when extracting SIFT feature points,and PCA-based LDB descriptor is used as the feature expression.Finally,the feature matching strategy uses the traditional distance similarity measurement and the RANSAC error elimination algorithm.Experiment results show that the new matching algorithm not only effectively improves matching probability and matching accuracy,but also has strong adaptability for different landforms and strong anti-geometric distortion ability.An improved spatial resection positioning method is proposed to achieve accurate positioning of the aircraft by using the scene matching results.Firstly,based on the resection theory in photogrammetry and matching results,spatial collinear equations containing matching point pairs and pending information for position and angle are established.Then,in the process of solving the equations,the inertial navigation data is used as the initial value for iterative calculation,which effectively solves the problem of iterative solution for nonlinear equations.In the simulation experiment,the influence for positioning accuracy is deeply studied under five parameter perturbations: number of matching point pairs,zero drift of inertial navigation initial position,zero drift of inertial navigation initial attitude angle,ground point measurement error and initial attitude angle,which proves that the positioning model can obtain higher positioning accuracy under large measurement errors.A residual Kalman filtering model based on the fusion of inertial measurement information and matching positioning measurement information is established to further improve the navigation and positioning accuracy of the aircraft by using the filtered output to correct the inertial deviation.The model takes the residual of the INS data and the real data as the Kalman filter optimal estimation value,and takes the residual of the INS value and the matching positioning data as the state observation value,which obtains an optimal estimation for the spatial positioning information of the aircraft through an optimal estimation of the residual.The simulation results show that the model can not only effectively filter the initial zero drift of the inertial navigation,but also effectively suppress the violent fluctuation of the matching positioning error.It also has good robustness to inertial navigation initial position zero drift,inertial navigation initial attitude angle zero drift,matching error and ground control point measurement error.