Determination of the moisture change behavior of cross-laminated timber using an optical flow based computer vision technique

Abstract

Although cross-laminated timber (CLT) offers greatly improved directional stability against moisture changes compared to lumber, some layup dependent directional differences still remain. Furthermore, even under a purely homogeneous distributed moisture change strong deformations appear along the boundaries, which show a decrease of swelling/shrinkage towards the inside of CLT panels. Metrological determination of this behavior is still a challenging task and involves longlasting moisture content conditioning and typically manual measurements. This limits the amount of measurable data-points and thus the gain-able insights. We apply a recently introduced computer vision technique based on optical flow from scan images to measure surface deformation fields of various CLT specimens with different layups. This allows us to measure the change of average differential swelling and shrinkage coefficients throughout the cross section and visualize them as curves with high resolution. We gain measurements for each image pixel and demonstrate good matching to previously published manual single-point measurements. Furthermore, we analyze various specimens specifically built to allow for investigations of the aforementioned boundary effects. Using the computer vision approach we are able to show how the combination of homogeneous deformations and boundary effects leads to the resulting deformations observable with manual methods.