Research On The Key Technology Of Target Position Based On Feature Image Matching

With the increasing demand for national defense and public security,the performance requirements for the ground target positioning are also getting higher and higher.The airborne photoelectric equipment is an effective means to achieve precise ground target positioning.The performance of target positioning determines the success of the target accurate monitoring and combat missions.The traditional methods of the airborne photoelectric equipment target positioning have many error sources and are vulnerable to environmental interference.The image-based target positioning technology is not susceptible to signal interference and has high positioning accuracy,so it has been the mainstream of the research.The article regerds the airborne photoelectric equipment as the research object,and analyzes and researches the key technologies of target positioning based on feature image matching from three parts including the overall target positioning scheme,homologous large perspective image matching,heterogeneous large perspective image matching and the target positioning based on feature image matching.First of all,the thesis analyzes the advantages and disadvantages of the traditional methods of the airborne photoelectric equipment target positioning,researches the overall scheme of the target positioning based on feature image matching,and then designs the theory architecture.At the same time,the basic theory of feature image matching is analyzed.The thesis focuses on the analysis of the performance of classical edge feature extraction,point feature extraction and description methods,which lays a good foundation for the following researches on the homologous large perspective image matching and heterogeneous large perspective image matching algorithms.In order to solve the problem that it is hard to match images with large angle difference and meanwhile improve the operation speed and reliability of the algorithm,a PORB large perspective image matching algorithm is proposed.This algorithm uses ORB fast matching algorithm as the basic algorithm,eliminates the image distortion through exhaustive sampling of the image perspective simulation,and realizes the matching of images with large angle difference.Aiming at the shortcoming of exhaustive sampling PORB algorithm uses,a fast PORB large perspective image matching algorithm based on homography matrix is proposed.By setting pre-judgment,the number of loop simulation is reduced and the speed and reliability of the algorithm are improved.In order to solve the shortcomings of poor matching performance between infrared and visible images because of their large difference in gray level,an improved ORB heterogeneous image matching algorithm based on adaptive Canny edge feature is proposed.By point feature matching in the edge image,infrared and visible image matching is achieved.This method is difficult to adapt to the case of large perspective image matching.Therefore,a new PORB heterogeneous large perspective image matching algorithm based on adaptive Canny edge feature is also proposed.With reference to the fast PORB algorithm,the heterosexual large perspective image matching between infrared and visible images is achieved by the perspective simulation and pre-judge setting.Finally,in order to realize the target positioning of the airborne photoelectric equipment,a target positioning method based on feature image matching is presented.According to the result of image matching,the correspondence between the target positions in the two images is obtained,and then the method achieves the target high-precision positioning.At the same time,a set of visual simulation verification platform is designed and implemented using the existing conditions in the laboratory.The visual simulation platform integrates flight simulation,image simulation,image matching and target localization,and directly reflects the superiority of the target positioning method in the thesis.