Quality inspection method for splicing welds of PHC piles based on line structured light

The accuracy of extracting the centerline of weld seam laser directly affects the accuracy of weld seam inspection detec- tion. However, in practical conditions, due to significant surface undulations of PHC piles, reflections on the construction site, and insufficient laser power, existing algorithms are easily affected by noise . To address these issues, this paper proposes a method combining an internally advanced gray centroid method to extract the initial center point. To meet industrial measurement accuracy requirements, principal component analysis is used to solve the normal of the initial center point, and secondary optimization of the center point is performed along the normal direction. Feature point extraction algorithms are integrated to detect feature points of the weld seam laser stripes, which are then transformed into the camera coordinate system for weld seam formation quality detection. Experimental results show that, compared to manual measurement, this method offers higher accuracy and shorter time consumption, meeting the real-time requirements of industrial applications.