Solution Method Of Space Target Position And Orientation Based On Binocular 3D Point Cloud

With the increasing development of space exploration activities,On-orbit servicing and other space activities of the faulty spacecraft have become an important subject to the development of aerospace technology.However,the pose measurement technology of the spatial target is the key factor to solve these problems.Therefore,it is very important to study the short distance pose measurement of space target.In this paper,the spatial target is the main object of study;this article puts forward a pose measurement method of the spatial target based on the binocular vision after looking up the current research of the on-orbit services of the non-cooperation target and the spatial target visual measurement technology.This method first applies the principle of the binocular vision measurement system to reconstruct the three-dimensional point cloud of the space object,which involves the image correction based on the epipolar geometry theory and the stereo matching algorithm based on the image gray scale correlation normalization(NCC),and then estimate the pose of the spatial target with the method of point cloud registration which based on the classical ICP algorithm.In this paper,the image model and the aberration correction model of the camera are introduced.And the internal and external parameters of the binocular system are obtained by using the Halcon calibration toolbox based on the plane target camera calibration method.In terms of 3D point cloud reconstruction,this paper first uses the binocular stereoscopic image correction method based on the epipolar geometry to correct the spatial target image acquired by the measurement system,and then applies the NCC stereo matching algorithm based on the image gray scale to the corrected image to obtain the point cloud image by correspondence matching,and then calculate the three-dimensional coordinates of the point cloud with the parallax theory.Finally achieve the 3D point cloud model of the target.In terms of point cloud registration,this paper uses the classic two-step point cloud registration method,that is,the initial rough registration and precise registration afterwards.Firstly,the target point cloud obtained by 3D reconstruction and the known reference point cloud are initially registered by the principal component analysis(PCA),so that the two points cloud roughly coincide;then the two points cloud after initial registration are applied with the improved ICP algorithm by accurate registration to obtain the pose of the space object.For the pose solution method of the spatial target mentioned above,this article uses the 3dsMax rendering design software which simulates the process of the binocular system to obtain the simulation images for 3D point cloud reconstruction.Besides,this article achieved the goal of the reconstruction of 3D point cloud model and pose solution with the assistance of the machine vision software Halcon which developed by MvTec company in Germany.Through a large number of simulation experiments,it proves the validity and robustness of the algorithm,and obtains a high precision pose measurement results in short distance.