A Comparative Study on Pose Estimation for Monocular Vision and Binocular Vision Without Modeling

摘要

Pose estimation algorithms with regard to both monocular vision and binocular one have been studied respectively since about twenty years ago, however, few researchers considered comparing two representative methods for the sake of actual application. In this paper, considering the factual model, we have done a comparative study about these two methods quantitatively. The statistical result of calculated error based on simulation experiments denotes that the model-based monocular vision behaves more outstanding than the binocular vision on the computational accuracy and robust inasmuch the monocular vision introduces the known object model in the process of computing the coordinates of feature points. The simulated result has certain guided significance for choosing a suitable pose estimation algorithm in actual engineering projects.