A Self-calibration Method For Binocular Vision System Of Space Robot In Complex Light Environment

The effective calibration of the robot vision system can realize the accurate positioning of the movement and the operation target.In the process of launch and launch,the robot vision system may fail,or according to the actual mission needs of the camera focal length,aperture and the relative position between cameras to re-adjust,which requires on the station to re-calibration.It is currently the main method to calibrate the target.However,considering that the robot needs to be calibrated outside the cabin,it needs to install the target on the cabin,and the surface of the cabin is usually curved.If the target is installed,it needs additional design device.Which is difficult to be implemented in the space station.Therefore,this paper proposes an autonomous calibration method based on orthogonal parallel lines.This method does not need to use the target,and can realize effective calibration under complicated illumination environment of the space station,which satisfies the robot vision system positioning accuracy requirements.This paper focuses on the autonomous calibration of space robot binocular vision system under complex illumination environment.The main research contents and achievements are as follows:First,the complex lighting environment of the space station will bring a lot of mixed noise to the visual system,which brings new challenges to the binocular vision system.In this paper,a Kalman scaled image filtering algorithm based on maximum correlation entropy is proposed to solve the dynamic noise problem caused by illumination mutation,lens shaking and target occlusion in the space station,which provides the basis for the subsequent parallel line identification and extraction.Secondly,in order to improve the accuracy of parallel line recognition and extraction in space station,this paper proposes an image enhancement and parallel line identification and extraction scheme.Firstly,the histogram equalization and Retinex enhancement are used to optimize the static noise caused by light exposure,lack and shadows in the image,and then the proposed algorithm is used to filter the image in the image based on the maximum correlation entropy Kalman calibration algorithm.Finally,combined with Canny operator and Hough transform to extract the orthogonal parallel lines in the field of view,the accurate identification and extraction of the parallel lines in the complex illumination environment of the space station is realized.Thirdly,in order to solve the problem of self-calibration of the binocular vision system,a geometric feature of orthogonal parallel lines is chosen as the parameter of the autonomous calibration algorithm.On the basis of realizing the algorithm,this paper will test the algorithm comprehensively,analyze the factors that affect the precision of the visual system,including the illumination environment,and improve the algorithm for the big factor,so as to realize this algorithm stably used in space stations.Fourthly,the experimental platform is designed to simulate the illumination environment of the space station,the comprehensive evaluation based on the target calibration method,the orthogonal calibration algorithm based on orthogonal parallel lines,the improved algorithm based on Kalman filter and the self-calibration based on the maximum correlation entropy Kalman filter Algorithm in the ideal condition,the change of light environment,the shaking of the lens and the occlusion of the target.The feasibility and validity of the proposed method are also verified by the proposed method of the binocular vision system.