Video Motion Magnification to Improve the Accuracy of Vision-Based Vibration Measurements
Eduardo Pérez, Emanuele Zappa
Abstract
Video motion magnification allows to turn a tiny movement recorded on video into an easily visible displacement. In this work, such techniques are employed to decrease the uncertainty of vision-based motion measurements. Previous works have already combined motion magnification with vision-based measurement techniques, such as digital image correlation. This present study analyzes its applicability on simpler measurement techniques: pattern matching and edge detection, analyzing the accuracy improvements when measuring on a motion-magnified video. In addition, the suitability for these purposes of recent motion magnification algorithms is tested, and the circumstances under which the magnification is beneficial are studied. As a result of the experiments, it is concluded that motion magnification can increase the motion measurement accuracy, especially for small displacements, which could be considered an initial step for future quantifications of its applicability range.