A Monocular Vision Pose Measurement Method Based on Harmony Search Algorithm

Abstract

In this paper,we proposed a monocular visual pose measurement method that combines the weighted iterative improved EPnP algorithm and the self-adaptive global best harmony search algorithm. First, we analyzed the process of EPnP algorithm to solve the target pose, combined with the depth of the marker point and the reprojection error to set the weight coefficient, and finally optimized the objective function by the self-adaptive global best harmony search algorithm. In order to verify the accuracy of the method proposed in this paper, we have carried out related experiments with a 6-DOF robot arm platform. The results shows that when the relative distance is 2M, the absolute error in the X and Y directions of the method proposed in this paper are less than 0.089cm, the maximum rotation errors in the X and Y directions are 0.236° and 0.211°, respectively. It is worth noting that in the depth change experiment, this algorithm can effectively reduce the influence of depth change, so that the absolute error in the Z direction is always less than 0.25cm. And the robustness in the Gaussian noise experiment is stronger, which can meet the actual measurement requirements in industry.