International Journal of Computer Vision
| Hi-index | 0.00 |
The problem of camera self-calibration and Euclidean reconstruction from image sequences is addressed in the paper. We propose a quasi-perspective projection model and apply the model to structure and motion factorization to estimate the focal lengths of the cameras. Then we optimize the camera parameters based on Kruppa constraints and recover the metric structure from factorization of the normalized tracking matrix. The novelty and contribution of the paper lies in two aspects. First, under the assumption that the camera is far away from the object with small rotations, we propose that the imaging process can be modeled by quasi-perspective projection. The model is more accurate than affine camera model since the projective depths are implicitly embedded. Second, we propose to calibrate a more general camera model with 5 intrinsic parameters, while previous factorization algorithm can only calibrate the focal lengths. We validate and evaluate the proposed method on many synthetic and real image sequences and show the improvements over existing solutions.