3D structure from uncalibrated image sequence

A new approach to recover 3D structure from the aerial uncalibrated image sequence is presented. Unlike previous methods, the method uses a factorization-based algorithm to recover 3D models' initial values and bundle adjustment to get accurate results. The focal lengths can be supposed to be the only unknown intrinsic parameters. The factorization-based method performs a projective reconstruction by using a linear factorization algorithm and upgrades the projective solution to a Euclidean one by enforcing some metric constraints. Bundle adjustment refines a visual reconstruction to produce jointly optimal structure and viewing parameter estimates when we get the ideal initial values from the factorization-based methods. Experiments results on synthetic and real image sequences verified the new approach's precise and efficiency.