Research on Feature Size Recognition and Measurement of High Voltage Cable Joints Based on 3D Reconstruction

This paper presents a method for recognizing and measuring feature sizes at critical process nodes to enhance the installation quality of high-voltage cable accessories and address the issues of imprecise, inefficient, and poorly defined inspection techniques. The method leverages three-dimensional reconstruction, employing a normal vector condition segmentation approach integrated with principal component analysis, which enables precise point cloud segmentation of critical nodes, such as stripping, conductor crimping, and stress cone installation. Additionally, it facilitates point cloud acquisition at installation sites and non-contact measurement of key feature dimensions. Experimental results demonstrate that the proposed method can rapidly and autonomously identify key feature sizes with a measurement error of less than 1%, meeting field application requirements.