Alternative weighted structure from motion under quasi-perspective projection model

The paper focuses on 3D structure and motion factorization from uncalibrated image sequences. An alternative weighted factorization algorithm is proposed to handle noisy image measurement. The novelty and main contributions of the paper are as follows: (i) A simple while accurate quasi-perspective projection model is adopted for structure and motion factorization; (ii) feature uncertainties are estimated directly from image reprojection residuals; (iii) an alternative weighted factorization scheme is proposed to reduce the influence of noisy data; and (iv) the proposed scheme is proved empirically to be more accurate than other factorization algorithms. The proposed approach is validated by extensive experiments on synthetic data and real images.