A Target Pose Processing Method based on Multi-Line Vector Object-Image Mapping Matching

In the practical application of middle& long-distance measurement, due to view occlusion, target movement or other reasons, it is difficult to extract stable feature points with the same feature points. In this paper, pure straight line vector is adopted to solve the pose processing problem. Based on th 3D-2D correspondence between linear vector object space and image space, this paper establishes the universal method of pose processing for axisymmetric rotating body targets and non-axisymmetric rotating body targets for the first time, and solves the ambiguity of monocular station pose processing; this method only uses a small amount of key line vector information, and has the advantages of no homonymous features, effective anti occlusion and strong processing flexibility in feature extraction; in terms of processing, the accuracy and stability are improved because the linear vector is based on pure direction solution, which excludes the usual coupling factors of position and pose.