Covariance Propagation for the Uncertainty Estimation in Stereo Vision

Stereo vision has emerged as an effective approach for contactless 3-D measurements. This paper presents a methodology for the propagation of the measurement uncertainty through stereo-vision calibration and measurement algorithms, according to international standards. The proposed procedure is compared to the reference statistical procedure for uncertainty evaluation, and the results are presented and discussed. The results proved that the proposed procedure was able to estimate the standard uncertainty even at considerable working distances from the calibrated volume with negligible processing times and far more accurately than the theoretical simplified models commonly adopted.