New Robust Bayesian Method for the F-Matrix Estimation

摘要

In this paper a new robust method is proposed for the fundamental matrix211u001eestimation. The method is based on the optimal Bayesian weighting of the interest 211u001epoints by their probability to be relevant. The method is evaluated with the 211u001eaffine F-matrix estimation problem with both synthetic and real data, but its use 211u001ewith the full perspective projection model is also discussed.