Research On Key Technologies Of Binocular Vision Measurement Based On Space Robotic Arms

Based on the background of the space robotic arm,this paper studies the binocular vision technology which is widely used in the space robotic arm.The space robotic arm is a space equipment which integrates multiple subjects including machine,vision,electron and so on,and it is of importance to the space station.According to the technology of vision,space robotic arm enables all kings of operations,such as docking,maintenance and removal of orbital debris.With the rapid development of vision technology,binocular vision technology has attracted more attention in the space field.The technology perceives external scenes by simulating human eyes and it is an important part of computer vision.With its advantages of high efficiency,simple structure and low cost,this technology is widely used in the fields of virtual reality,medical imaging and industrial detection.The research on position measurement,attitude estimation and 3D reconstruction is of great significance,which is based on the binocular vision technology.With the increasing demand,binocular vision system is facing new challenge in improving accuracy.In response to this challenge,this paper studies the binocular vision system,camera calibration and stereo matching,and the works completed include the following aspects:1.This paper studies the binocular vision system and analyses the camera calibration,image correction,the principle of binocular matching and 3D reconstruction.Based on the model of perspective projection,this paper studies the geometrical relationship of the three coordinate systems(space coordinate,camera coordinate and image coordinate).2.At the stage of camera calibration,this paper analyses the principles of calibration.Camera calibration is one of the key steps in binocular vision system and the accuracy of calibration results directly affects the subsequent stereo matching and 3D reconstruction.Based on the existing calibrations methods,this paper mainly analyses the calibration method proposed by Zhang Zhengyou.This method is between traditional calibration method and self-calibration method and combines their advantages.Secondly,this paper carries on the Zhang Zhengyou calibration experiment and the experimental results show that the method has high calibration accuracy which meets the needs of following works.Finally,according to the internal and external parameters obtained by the calibration,this paper accomplishes the work of correcting images.3.This paper focuses on the research of stereo matching methods from three aspects including initial matching cost calculation,cost aggregation,disparity calculation and optimization.At the stage of initial matching cost calculation,this paper proposes a similarity measure which combines the improved Census transform,mutual information and gradient information;At the stage of cost aggregation,the idea of adaptive weight is introduced on the traditional guided filtering,so that the regularization parameters in different windows can be adjusted adaptively.At the stage of disparity calculation and optimization,the strategy of WTA(Winner Takes All)is used to select the initial disparity and optimizing disparity by reliability criterion and left-right consistency detection.Finally,this paper makes an experiment about stereo matching.Compared with other matching methods,the results show that the proposed method has certain advantages.4.This paper studies the principle of 3D reconstruction.The key of reconstruction is containing the coordinates of space points,and the visual reconstruction of 3D scene can be realized by Delaunay triangulation and texture mapping.The experimental results show that the reconstructed image can reflect the spatial information of the original scene.