Compute the relative orientation between two cameras given image point correspondences and known camera parameters and reconstruct 3D space points.

For a stereo configuration with known camera parameters the geometric relation between the two images is defined by the relative pose. The operator vector_to_rel_pose computes the relative pose from in general at least six point correspondences in the image pair. RelPose indicates the relative pose of camera 1 with respect to camera 2 (see create_pose for more information about poses and their representations.). This is in accordance with the explicit calibration of a stereo setup using the operator binocular_calibration. Now, let R,t be the rotation and translation of the relative pose. Then, the essential matrix E is defined as E=([t]_x R)^T, where [t]_x denotes the 3x3 skew-symmetric matrix realising the cross product with the vector t. The pose can be determined from the epipolar constraint:

            T
      / X2 \            T   / X1 \                     /   0  -t_z  t_y \
      | Y2 | * ([t]_x R)  * | Y1 | = 0  where  [t]_x = |  t_z   0  -t_x |  .
      \  1 /                \  1 /                     \ -t_y  t_x   0  /

Note, that the essential matrix is a projective entity and thus is defined up to a scaling factor. From this follows that the translation vector of the relative pose can only be determined up to scale too. In fact, the computed translation vector will always be normalized to unit length. As a consequence, a threedimensional reconstruction of the scene, here in terms of points given by their coordinates (X,Y,Z), can be carried out only up to a single global scaling factor. If the absolute 3D coordinates of the reconstruction are to be achieved the unknown scaling factor can be computed from a gauge, which has to be visible in both images. For example, a simple gauge can be given by any known distance between points in the scene.

The operator vector_to_rel_pose is designed to deal with a camera model that includes lens distortions. This is in constrast to the operator vector_to_essential_matrix, which encompasses only straight line preserving cameras. The camera parameters are passed by the arguments CamPar1, CamPar2. The 3D direction vectors (X1,Y1,1) and (X2,Y2,1) are calculated from the point coordinates (Rows1,Cols1) and (Rows2,Cols2) by inverting the process of projection (see camera_calibration). The point correspondences are typically determined by applying the operator match_rel_pose_ransac.

The parameter Method decides whether the relative orientation between the cameras is of a special type and which algorithm is to be applied for its computation. If Method is either 'normalized_dlt' or 'gold_standard' the relative orientation is arbitrary. Choosing 'trans_normalized_dlt' or 'trans_gold_standard' means that the relative motion between the cameras is a pure translation. The typical application for this special motion case is the scenario of a single fixed camera looking onto a moving conveyor belt. In this case the minimum required number of corresponding points is just two instead of six in the general case.

The relative pose is computed by a linear algorithm if 'normalized_dlt' or 'trans_normalized_dlt' is chosen. With 'gold_standard' or 'trans_gold_standard' the algorithm gives a statistically optimal result. Here, 'normalized_dlt' and 'gold_standard' stand for direct-linear-transformation and gold-standard-algorithm respectively. All methods return the coordinates (X,Y,Z) of the reconstructed 3D points. The optimal methods also return the covariances of the 3D points in CovXYZ. Let n be the number of points then the 3x3 covariance matrices are concatenated and stored in a tuple of length 9n. Additionally, the optimal methods return the 6x6 covariance matrix of the pose CovRelPose.

If an optimal gold-standard-algorithm is chosen the covariances of the image points (CovRR1, CovRC1, CovCC1, CovRR2, CovRC2, CovCC2) can be incorporated in the computation. They can be provided for example by the operator points_foerstner. If the point covariances are unknown, which is the default, empty tuples are input. In this case the optimization algorithm internally assumes uniform and equal covariances for all points.

The value Error indicates the overall quality of the optimization process and is the root-mean-square euclidian distance in pixels between the points and their corresponding epipolar lines.

For the operator vector_to_rel_pose a special configuration of scene points and cameras exists: if all 3D points lie in a single plane and additionally are all closer to one of the two cameras then the solution in the relative pose is not unique but twofold. As a consequence both solutions are computed and returned by the operator. This means that the output parameters are of double length and the values of the second solution are simply concatenated behind the values of the first one. This is valid for all output parameters but Error, which indicates the overall error of both solutions.

vector_to_rel_pose is reentrant and processed without parallelization.

match_rel_pose_ransac

gen_binocular_rectification_map, rel_pose_to_fundamental_matrix

vector_to_essential_matrix, vector_to_fundamental_matrix, binocular_calibration

camera_calibration

Richard Hartley, Andrew Zisserman: ``Multiple View Geometry in Computer Vision''; Cambridge University Press, Cambridge; 2003.

J.Chris McGlone (editor): ``Manual of Photogrammetry'' ; American Society for Photogrammetry and Remote Sensing ; 2004.

3D Metrology