Metrological performances of camera self-calibration techniques

Self-calibration algorithms are useful tools for different applications, whenever a geometrically known target cannot be used. This paper considers self-calibration algorithms for image-based measurement systems. After a brief theoretical introduction, a solution for the correction of the lens distortion is presented based on a genetic algorithm. A comparison of the performances achieved with different solutions and different values of the parameters is performed through several numerical tests, including the simulation of random errors in the localization of features, and a procedure for the evaluation of the performances achievable after a 3-D reconstruction of points is presented.