Binocular structured light-based three-dimensional measurement for refractory brick

To address the three-dimensional (3D) size measurement of low efficiency, incomplete geometric feature in point cloud surface reconstruction, and fitting deviation or stability in oriented bounding box, an improved 3D reconstruction method based on binocular structured light is proposed for accurate and rapid 3D size measurement of refractory brick. First, the image acquisition system used a pair of synchronised cameras and a digital projector to capture structured light images of refractory brick. Second, absolute phase map-based stereo matching is used to improve the matching accuracy and efficiency for weak and irregular textures on the refractory brick surface. Subsequently, a multi-directional fast marching (MDFM) algorithm is proposed for surface reconstruction from the 3D point cloud of refractory brick. Finally, a gradient descent optimised oriented bounding box (GDO-OBB) algorithm is developed to improve the bounding box for 3D size measurement. Experimental results show that the relative error rate e_R, standard deviation of repeatability accuracy σ_RepA, fluctuation coefficient τ, and centroid deviation rate Δ_CS between the true and measurement values are 2.50, 0.57, 508.18, and 0.051, respectively. This study shows that high-accuracy 3D size measurement of refractory brick can be achieved by binocular structured light-based 3D reconstruction, which would be useful for subsequent brick grasping and handling in real-world intelligent construction scenarios.