Accuracy improvement in area-based matching for structural displacement measurements

Abstract

Measuring and monitoring of structure deformations such as beams have an essential role in civil structural analysis. Measurements obtained pertaining to their displacements, among others, provide the information needed for the studies on material behaviours and structural designs. These measurements can also provide important indicators regarding to their failures. Under controlled laboratory conditions, these displacements can be determined using, for instance, high precision Linear Voltage Differential Transducers (LVDT). The high precision capabilities of these sensors make them suitable for structural deflection experiments. However, these LVDT sensors face a number of major drawbacks, such as, the sensors may be subjected to movement or damaged during the experiment, and the points measured are at pre-determined locations. In other words, displacements can only be measured at points where the LVDTs sensors are fixed. In addition, when large numbers of points of displacement are required or desired, the use of these sensors becomes prohibitively expensive and laborious. Whilst various researchers have used digital close range photogrammetry and the area-based matching approach in determining movements but work on the use of more than two images and surface models has not been reported. Therefore, this study proposes a revised method of precisely determining the displacements of structures using a multi-image area-based matching approach that uses surface models, i.e. a non-contact method. Experiments on beams under loading were performed under laboratory conditions. A series of multiple digital images were captured simultaneously using three digital single-lens reflex cameras throughout the experiments. The beam’s vertical displacements obtained from the proposed method were then validated by comparing against those obtained from the LVDTs. The results indicate that the mean differences between the displacement obtained from the proposed method and LVDTs are less than 0.5mm.The ttest conducted with a confidence level of 5% revealed that the differences between the two sets of results are not significant. It can be concluded that the use of multi-image area-based image matching using surface models is capable of measuring displacements and be used as an additional approach that complements the traditional methods in beam displacement measurements.