Soft methods for bounding the uncertainty of stereo calibration and triangulation

3D triangulation with stereo vision is subjected to different sources of uncertainty that propagate to the final result. In previous work it has been shown that the tolerance of the triangulation can be delimited by extending calculations to interval data. Notwithstanding this, in this paper it will be shown that this kind of estimation may produce too loose bounds in many practical circumstances. A Soft Computing based algorithm is proposed instead that improves prior methods. A sample of combinations of calibration points extrema is used to generate a set of camera calibration parameters. A fuzzy set is fitted to this last set of data, that measures the compatibility of 3D coordinates with a pair of 2D matched images.