Estimating small structural motions from multi-view video measurement

The phase-based estimation, as one type of target-free methods, can estimate small structural motions from video data. However, in the two-dimensional (2D) domain, this method is prone to noisy errors because of the ill-posed problem, and its measurement range is limited due to the periodicity of the phase variation. Additionally, it cannot estimate three-dimensional (3D) structural motions. To estimate 3D small structural motions more accurately, this paper proposed a novel multi-view video measurement method including 2 steps, namely weighted phase unwrapping and lq-norm minimization. The weighted phase unwrapping is to measure 2D small structural motions, in which the unweighted and weighted least-square formulations are established, and iter-atively solved by the fast transforms. To project 2D structural motions into the 3D domain, the lq-norm multi -view minimization with the corresponding reweighted algorithm is developed. The proposed method was applied in the experiment of a 3-storey reinforced concrete (RC) structure under earthquake loads. Its validity was verified by the more accurate estimation of the structural motions from multi-view video measurement.