Monocular pose measurement method based on circle and line features

Abstract

In this paper, a new method based on circle and linear projection is proposed. In a straight line and a circle on the target space as the main object recognition. In the case of knowing the radius of the space circle and coordinate of a point on the space line: First of all, we can use the space circle projection to solve all possible solutions, there are two groups. Secondly, determine the rotation angle of space circle around normal vector in the two groups by using the relative position relation of space straight line and circle. Then, we can get the rotation relation between the object coordinate system and the camera coordinate system and get the coordinate representation of circle and straight line in the camera coordinate system in two groups of solutions. Finally, project the point on the space line onto the image plane in two groups of solutions. In the correct solution, the distance between the projection point and the projection line is small. The experimental results show that the method can find the correct solution in all solutions and determine the rotation angle of the space circle around the normal vector. The method can get the correct result under the condition of noise, so the method has strong robustness.

Elliptic shape prior for object 2D-3D pose estimation using circular feature

“…The existing algorithms detect elliptic from 2D image, and the 3D pose of the circular can be extracted from single image using the inverse projection model of the calibrated camera (see Fig. 1b) [3][4][5]. These methods successfully applied to pose estimation of underwater dock [6].…”