In computer vision, the camera calibration is a process needed when the recovery of some information, such as angles and distances, is desired. The present work deals with the problem of camera calibration using one-dimensional patterns. Nowadays, this problem only has a solution if some restrictions to the pattern's movement are imposed or if some angles of the cameras are known in advance. However, a different approach can be applied if, instead of only one camera, a stereo system is considered. In that case, the calibration is possible with a one-dimensional pattern that executes an unknown and unrestricted movement, even without any previous information concerning the cameras. Such method is based on the estimation of a transform which, after the estimation of the system's fundamental matrix, allows updating a projective calibration into a Euclidean calibration. Experiments using both real and synthetic images validate the new method and demonstrate that its accuracy is comparable to other well known calibration methods in the literature.
Camera Calibration; Calibration Pattern; 1D Object; Stereo Calibration
