Monocular 6D pose estimation method based on regular octahedron

The invention discloses a monocular 6D pose estimation method based on a regular octahedron. According to the method, the ArUco code and the regular octahedron are creatively combined, pose recognition is expanded from two-dimensional label recognition to three-dimensional space body recognition, pose recognition performance is greatly improved, a monocular camera is used for collecting a regular octahedron image marked by the ArUco code, a pose matrix of the ArUco code is calculated through a PnP algorithm. The pose of the measured object is indirectly calculated according to the relative pose matrix of the ArUco code on each surface of the regular octahedron in the regular octahedron coordinate system and the pose matrix of the regular octahedron of the current frame relative to the regular octahedron of the initial frame, which is calculated based on the theory that the larger the area of the detected ArUco code is, the more credible the area of the detected ArUco code is. Areas of obtained ArUco codes are