Abstract

The pinhole imaging model assumes that all the projecting rays intersect at the effective pinhole point. The restriction of this camera model will result in low accuracy of vision-based pose estimation. Furthermore, the camera calibration gets the global optimal solution of intrinsic and extrinsic parameters, which will affect the calibration accuracy of intrinsic parameters and then pose estimation accuracy is also affected. In this paper we completed the pose estimation with four coplanar feature points via an incident-ray-based camera model. Two parallel planes are used to perform the ray-pixel mapping. The projecting ray could be defined by its two intersections with the two planes. In this way, the camera is parametrized by the image plane and the two parallel planes. Based on this camera model, through four coplanar feature points, the object pose could be estimated. The whole process consists of both a non-iterative step and iterative step. In the non-iterative step, the initial estimation of the pose is obtained and then it is refined in the iterative step. Experiment results on pose estimation by the proposed algorithm and the other existing algorithms demonstrate the superiority of our algorithm.

© 2021 Optical Society of America

PDF Article

References

You do not have subscription access to this journal. Citation lists with outbound citation links are available to subscribers only. You may subscribe either as an Optica member, or as an authorized user of your institution.

Contact your librarian or system administrator
or
Login to access Optica Member Subscription

Cited By

You do not have subscription access to this journal. Cited by links are available to subscribers only. You may subscribe either as an Optica member, or as an authorized user of your institution.

Contact your librarian or system administrator
or
Login to access Optica Member Subscription