Research On UAV Visual Navigation And Location Technology Based On Scene Matching

The autonomous navigation of UAVs during high-altitude flight has a critical impact on the degree of automation.The visual navigation positioning technology based on scene matching does not rely on external systems compared to GPS navigation,and there is no cumulative error compared to inertial navigation.This technology is a navigation and positioning technology with high autonomy and high precision.It plays an increasingly important role in the field of navigation and positioning,and is also the frontier of the visual field.The scene matching algorithm achieves geolocation by locating the position of the real-time map in the reference map.In this paper,we compare and contrast different scene matching algorithms,and design and focus on a kind of scene-based scene matching algorithm based on SURF features,and optimize the algorithm from the perspectives of speed,precision,autonomy and rapid relocation.Firstly,this paper explores the application background of UAV visual navigation and positioning technology based on scene matching,clarifies the research significance,analyzes the development status of scene matching algorithm,and introduces the main research contents and research schemes of this paper.In addition,the definition,basic composition,basic framework and common classification of the scene matching problem are also given.Then,this paper clarifies the test scheme of scene matching algorithm,and studies the acquisition method of benchmark map,realizes the method of acquiring multiresolution satellite map in any range from web map service platform.In the test scheme,this paper uses Google Maps tile splicing to form a test reference map,to completes the test comparison of the scene matching algorithm of edge features and the scene matching algorithm of SURF features,and proves that the latter is superior in accuracy and speed,so the latter is the object of focus research and optimization.Finally,this paper designs four optimization schemes for multiple shortcomings of the algorithm..Aiming at the possibility of matching positioning failure,the rationality judgment and failure relocation scheme are designed,and the rational judgment can give hints when the matching is wrong.The relocation scheme gives four recommended methods to achieve relocation;In terms of precision optimization,the influence of Hessian matrix determinant threshold on the accuracy of the algorithm is explored,the matching precision is improved to over 90%,and the recommended value of Hessian matrix determinant threshold is given.For the optimization of speed,using historical information and UAV performance parameters,the matching speed is greatly increased from 1 fps to more than 30 fps under the test conditions in this paper.