Research On The Key Technology For Image Matching Aided Navigation System Based On Image Sequence

Navigation system with high autonomy and high accuracy can be constructed by fusing the visual information with the inertial navigation system.At present,it is a research hotspot to find stable information fusion algorithms and image matching algorithms with both high precision and real-time.Aiming at the need of practical engineering applications,this paper studies the key technology of image matching aided navigation system using the combined methods of theoretical study,system modeling and simulation.The main work is as follows:In the matching process between the image sequence and the reference map,as the SIFT algorithm is time consuming in search process,we propose a hierarchical search strategy.Firstly,high-contrast points are used to obtain rough position information.Then the reliable center of rough space position will be obtained by using DBSCAN method.Finally,carry out an exact match and delete the wrong points.This method can effectively improve the efficiency of the SIFT algorithm.The velocity information of the aircraft can be obtained by using image sequence matching.This paper uses the mean Hausdorff distance to estimating the motion between frames.In the anti-rotation requirement,this paper presents a method based on Harris corner and mean Hausdorff distance.Compared to other Hausdorff distance algorithm based on edge,the characters of anti-rotation,real-time and accuracy have been improved.Since there is a cumulative error in INS for long time work,in this paper,a mathematical model of fusion filter is established,which takes INS as the core and is also supported by image matching module.Aiming at the problem of information delay in image module,this paper presents a delay correction algorithm based on the INS information increment.Flight simulation experiment shows that the navigation accuracy is higher than the traditional Kalman filter algorithm and the correction algorithm based on curve fitting.Finally,simulation software of image matching aided navigation based on image sequence is developed to provide a visual test platform for the proposed image matching algorithms and filtering algorithm.And the validity of the proposed algorithms is also verified.