Relative orientation of multiple images using projective singular correlation
Ilkka Niini · Proceedings of SPIE, the International Society for Optical Engineering/Proceedings of SPIE · 1994
A relative orientation method of multiple images with unknown interior orientations is presented. The method is based on the projective singular correlation between the images and on the relations between the singular correlation, relative orientation, epipolar geometry, and interior orientation. The method does not require any control points as a reference, but at least three images taken with the same camera from distinct standpoints are required. The computations are linear in principle, and only the solution of the interior orientation is iterative and requires approximate values which also can be obtained linearly from a set of three images taken from the same position in space. The relative orientation method has four steps. First, all singular correlations between the images are computed and the epipoles are extracted from the correlations. Second, the interior orientations of the images are computed. Third, the relative orientations of the images are computed. Finally, the model coordinates of the object are obtained using the relative orientation parameters. The method can be applied for 3-D object reconstruction in a CAD/CAM environment.