The movement of a pedestrian bridge retrieved by means of image processing technique has been analysed in this paper. An optical target has been attached to the deck and its oscillation has been tracked with fast cameras. The movement of the bridge has also been measured with a radar interferometer and this result has been taken as the reference signal. Using these data, a parametric study of the errors introduced by the image-based methods has been performed. The influence of some variables in the measurement error such as the distance to the target, the image size, the type of camera or the movement amplitude has been analysed for four different distances, and two types of excitations. Results show that the relative error decreases with the amplitude and the target diameter and it increases with the target distance. Additionally, the maximum relative error obtained in most of the analysed cases is below 10%.