Research On Key Technologies Of Monocular Visual Pose Measurement In Complex Background

Vision measurement is a comprehensive discipline which involves optics,optoelectronics,image processing,pattern recognition and computer vision.With the continuously exploring of outer space,the rendezvous and docking and the mechanical arms to capture the object accurately become more and more frequent,the real-time and high precision measurement of the position and orientation of the target aircraft is an indispensable part of these missions.The great advantages of the monocular vision pose measurement are simple construction,lower cost,strong real-time capability,stable performance,high measuring accuracy compared with other methods,so the monocular vision pose measurement is widely applied in aerospace domain.Therefore makes a deeply research on monocular vision pose measurement has important significance for space exploration.This article focuses on the key technology research on monocular vision pose measurement under complex background,and designed a ground testing system based on monocular vision pose measurement.The main research works of this paper are as follows:(1)The camera imaging model and calibration technology has been discussed,based on the traditional liner camera model,mainly studied the camera lens distortion model,and introduced the model into the Zhang's camera calibration method,precise measure the intrinsic parameters of the fixed-focus camera used in the experiment.(2)The extraction of the space target features including the centroid location of light spot and the circle detect.Based on the study of the traditional centroid location algorithm,developed the weighted gray centroid and the Gaussian fitting method by changing the selection of light spot region,the simulation results show that the developed algorithm achieves better results than the traditional way.Combined the figure of satellite,we made a model which has the same size of the real satellite,take advantage of the satellite edge feature,we proposed four constraints involves the size of edge,the length-width ratio of the edge bounding rectangle,the line constraint,the ellipse area.The circle information on the surface of the satellite was detected effectively.(3)The space circle pose measurement based on monocular vision has been researched,making use of the reference points outside the circle to remove the false solution by comparing the distance of the projection of reference points and the imagine ellipse,and the simulation results show the proposed algorithm is feasible.(4)The ground testing system about monocular vision pose measurement has been set up,and the experiment results shows that the system has met the requirement precision,when the distance of camera and the target is 1200mm,the translation measurement error of target is 0.1mm/lOmm,the rotate error is 0.8 degree/5 degree.