Vision-Based Navigation Research For Aircraft Approching And Landing

At present, the methods for aircraft navigation when approaching andlanding mainly include global position system, inertial navigation system,instrument landing system and microwave landing system. In recent years,as the image processing technology gradually matures and computeroperation speed ascends， how to use visual information to acquire aircraftnavigation data becomes an essential research topic.According to the navigation requirements of aircraft approaching andlanding, this paper presents a method which uses image matchingtechnology and pinhole model to calculate the position, attitude and speedof the aircraft. This article’s main work includes:Matching of image local features: choosing SIFT as this paper’s imagematching algorithm according to the situation of aircraft approaching andlanding and the characteristic of the state of art algorithm. For theelimination of false image matching, this paper comes up with threeindependent models, while retaining high accuracy and strong robust of theoriginal algorithm, obtains remarkable acceleration effect, and enhance thespeed of the overall algorithm scheme to a certain extent.The camera model and calibration: summarize the pinhole cameramodel, calculate the intrinsic matrix.Position and attitude calculation and airport runway projection:calculating camera position and attitude using the EPnP algorithm, andprojecting the virtual runway onto the camera image according to theposition and attitude of the camera and runway geographic information,whose algorithm complexity is lower compared with runway detectionalgorithm based on visual features. Numerical simulation,3ds Max simulation and experimental resultsshow that, vision-based navigation can be used as one of the methods foraircraft navigation.