Research On Pose Estimation Method Of Flying Object Based On Contour Matching

Pose estimation is indispensable for aerial photography,aerial gravity,underwater terrain,gun attitude and direction aiming,vehicle/tank/plane direction and attitude and so on.The pose information is an important index to study the flight status and flight performance,providing an important basis for determining the reliability of flying targets and the hitting accuracy of flying weapons.It is of great significance to study the pose information of aircraft,such as the smooth operation of aircraft,the analysis of accident,the power of missile and the comfort of passengers.There are two main methods to determine pose information of flying objects: one method is to load GPS receivers and Inertial Measurement Unit in flying objects.Another approach is to use the optical images captured by photoelectric theodolite to determine pose information,which is adopted in this paper,however,there are some unfavorable factors in the imaging of photoelectric theodolite.The traditional methods of pose estimation would be ill conditioned.In this paper,a new method,contour matching method,is used in aircraft pose measurement.This method makes it possible to realize pose estimation of long distance moving target based on rigorous formula system of photogrammetry.Surveying adjustment method,from the view of mathematics,is a current method for optimal estimation of parameters.Classical least squares adjustment is to linearize the nonlinear function model by neglecting the higher order terms and solve the linearized model by linear least squares method.The determination of initial value and the degree of nonlinearity of the mathematical model are strictly required in this method,however,the mathematical model established by Generalized point Feedback Iteration method can not meet the requirements,which seriously affects the accuracy of surveying adjustment results.In this paper,several methods of measuring adjustment are compared and analyzed in detail through theoretical and experimental results.By means of numerical optimization,this paper presents an efficient solution for pose estimation of flying objects,which has quick convergence.In this paper,a complete pose estimation system is built based on the theory,and the accuracy and practicability of proposed algorithm can be proved by analysis of the pose estimation of simulation data.The main research conclusion received through the experiment is as follows: 1)Combined with the principle of photogrammetry and OpenGL,the imaging simulation of photoelectric theodolite is realized.Pose estimation of projection image based on contour matching and adjustment theory is established and completed.2)Through theoretical deduction,the optimization methods such as steepest descent method,Newton method,Gauss-Newton method and damped least square method are analyzed and compared.Combined with the mathematical model,the iterative method of attitude determination based on damped least square is designed.The experimental results show that the aircraft pose parameters can be effectively computed.3)On the basis of the above mathematical model and solution method,a pose estimation software system for plane image contour matching is established.The simulation results show that the accuracy of yaw angle and pitch angle is better than 0.5°,and the accuracy of roll angle is better than 0.9°.