Floating wood volume determination using non-contact measurements and RANSAC method

Floating objects, especially larger ones, often accumulate on hydraulic structures, reducing their flow capacity and leading to overtopping, structural damage, and altered flow conditions, thereby increasing the risk of flooding and erosion. When floating objects combine with sediment transport at bridges or culverts, upstream water levels may rise, potentially undermining foundations and causing scouring. Understanding the quantity, characteristics, and movement of floating objects is essential for mitigating their adverse effects. To this end, a laboratory experiment was conducted, considering floating wood, one of the most common and voluminous river debris types. The study analysed four lumber samples and 65 wood logs. Measurements were performed using a two-dimensional (2D) industrial laser scanner and an industrial camera system. Both methods generate point clouds, from which individual volumes were calculated. Unlike previous approaches that estimated volumes from the object's above-water volume and material density, this study aimed for higher accuracy using the Random Sample Consensus (RANSAC) method. RANSAC, not previously used in similar studies, is a robust algorithm for fitting models to noisy data. It works by selecting random subsets of points and iteratively identifying the model with the best fit, maximising inliers. While computationally demanding and sensitive to parameter selection, RANSAC remains a reliable tool for model fitting and is widely applied in computer vision. The accuracy of calculated volumes was validated against reference values obtained using Archimedes' principle.