Circular target pose measurement method and device based on binocular inverse projection transformation

Abstract

The invention provides a circular target pose measurement method and device based on binocular inverse projection transformation. The method comprises the following steps: respectively carrying out elliptic contour feature detection on spatial circular target images shot by a left camera and a right camera to obtain corresponding elliptic contour pixel coordinates; inversely projecting the elliptic contour pixel coordinates corresponding to spatial circular target images shot by the left camera and the right camera into the space, and constructing a feature vector which starts from the optical center of the camera, passes through elliptic contour image plane coordinates and points to contour world coordinates of the spatial circular target ; according to the feature vectors, the initial pose information of the spatial circular target and the poses of the left camera and the right camera, acquiring world coordinates of the contour features of the reconstructed spatial circular target; constructing a residual function according to the difference between the radius of the reconstructed spatial circular target contour and the real radius of the circular target; and constructing an optimization model according to the residual function, and solving to obtain an optimal pose parameter of the space circular target, thereby realizing high-precision estimation of the pose of the long-distance small-size space circular target.

Description

technical field [0001] The invention relates to the technical fields of visual measurement and visual positioning, in particular to a method and device for measuring the pose of a circular target based on binocular inverse projection transformation. Background technique [0002] With the continuous development of target recognition, tracking, positioning and other technologies, the requirements for target positioning accuracy are getting higher and higher. In target positioning, by establishing a monocular or binocular vision positioning system, using circular features or adding a small amount of feature information The positioning of the target can be realized, and due to the symmetry of the circle itself, some random errors can be balanced, so that the positioning accuracy is generally better than the positioning method based on point features or line features. In the fields of industry, aerospace, medical and other fields, if the system design of visual perception and mea...

AI Technical Summary

Problems solved by technology

However, the solution process is too cumbersome. In the interaction between the image and the three-dimensional space circular target, the advantages of the binocular vision system itself have not been fully utilized, making the solution steps often too cumbersome. More importantly, no matter the above Whether the algorithm is based on monocular vision or binocular vision, it usually cannot give an accurate pose solution when the size of the circular target is much smaller than the measurement distance.

Images