get_circle_pose — Determine the 3D pose of a circle from its perspective 2D projection.
Each ellipse in an image can be interpreted as the perspective projection of a circle into the image. In fact, for a given radius of the circle, there exist two differently oriented circles in 3D that result in the same projection. get_circle_pose determines the 3D positions and orientations of these two circles. First, each Contour is approximated by an ellipse. Then, based on the internal camera parameters (CameraParam) and the radius of the circle in 3D (Radius), the 3D positions and orientations (Pose1,Pose2) are determined in camera coordinates.
Depending on the value of the parameter OutputType, the position and orientation is returned as a 3D pose (OutputType = 'pose') or in the form of the center of the 3D circle and the normal vector of the plane in which the circle lies (OutputType = 'center_normal'). In the former case, the angle for the rotation around the z axis is set to zero, because it cannot be determined. In the latter case, the first three elements of the output parameters Pose1 and Pose2 contain the position of the center of the circle. The following three elements contain the normal vector. The normal vectors are normalized and oriented such that they point away from the optical center which is the origin of the camera coordinate system. If OutputType is set to 'center_normal', the output parameters Pose1 and Pose2 contain only six elements which describe the position and orientation of the circle instead of the seven elements of the 3D pose that are returned if OutputType is set to 'pose'.
If more than one contour is passed in Contour, Radius must either contain a tuple that contains a value for each contour or only one value which is then used for all contours. The resulting positions and orientations are stored one after another in Pose1 and Pose2, i.e., Pose1 and Pose2 contain first the pose or the position and the normal vector of the first contour, followed by the respective values for the second contour and so on.
The accuracy of the determined poses depends heavily on the accuracy of the extracted contours. The extraction of curved edges using relatively large filter masks leads to a slightly shifted edge position. Edge extraction approaches that are based on the first derivative of the image function (e.g., edges_sub_pix) yield edges that are shifted towards the center of curvature, i.e., extracted ellipses will be slightly to small. Approaches that are based on the second derivative of the image function (laplace_of_gauss followed by zero_crossing_sub_pix) result in edges that are shifted away from the center of curvature, i.e., extracted ellipses will be slightly too large.
These effects increase with the curvature of the edge and with the size of the filter mask that is used for the edge extraction. Therefore, to achieve high accuracy, the ellipses should appear large in the image and the filter parameter should be chosen such that small filter masks are used (see info_edges).
Contours to be examined.
Internal camera parameters.
Number of elements: CameraParam == 8 || CameraParam == 10 || CameraParam == 12 || CameraParam == 14
Radius of the circle in object space.
Number of elements: Radius == Contour || Radius == 1
Restriction: Radius > 0.0
Type of output parameters.
Default value: 'pose'
List of values: 'center_normal', 'pose'
3D pose of the first circle.
Number of elements: Pose1 == 7 * Contour || Pose1 == 6 * Contour
3D pose of the second circle.
Number of elements: Pose2 == 7 * Contour || Pose2 == 6 * Contour
get_circle_pose returns 2 (H_MSG_TRUE) if all parameter values are correct and the position of the circle has been determined successfully. If necessary, an exception is raised.