DISCRETIZATION EFFECTS IN THE FUNDAMENTAL MATRIX COMPUTATION

A polyhedron represents the solution set of an approximate system modeling the epipolar constraints. We introduce a new robust approach for the computation of the fundamental matrix taking into account the intrinsic errors involved in the discretization process. The problem is modeled as an approximate equation system and reduced to a linear programming form. This approach is able to compute the solution set instead of trying to compute only a single vertex of the solution polyhedron as in previous approaches. Outliers are considered as sample point matches whose errors are much bigger that the expected uncertainty epsilon. We suggest ways to deal with outliers and present an analysis with experiments in synthetic images.