Linear Multi-View Reconstruction for Translating Cameras

This paper presents a linear multi view reconstruction algorithm for translating cameras with fixed internal parameters. The main advantages of this method are a) points and camera centers are computed simultaneously from one linear system containing all image data b) the allowance of arbitrary missing data. We show that the key to linearize the SFM problem is the infinite homography which comprises of the cameras’ calibration and rotation. This insight unifies reconstruction methods for calibrated cameras, e.g. Oliensis [9], and uncalibrated cameras, e.g. Rother-Carlsson [10]. A further contribution of this paper is the summary and comparison of different approaches to determine the infinite homography.