We consider reconstruction algorithms using points tracked over a sequence of (at least three) images, to estimate the positions of the cameras (motion parameters), the 3D coordinates (structure parameters), and the calibration matrix of the cameras (calibration parameters). We show how well maximum likelihood estimators perform, and how the choice of assumptions on the camera intrinsic parameters influences the precision of the estimator. We associate a Maximum Likelihood estimator to each type of assumptions, and derive analytically their covariance matrices, independently of any specific implementation. We verify that the obtained covariance matrices are realistic, and compare the relative performance of each type of estimator.
Keywords:
Uncalibrated 3D Reconstruction, Matched Image Features, ML estimation, Performance evaluation
