Three-dimensional Reconstruction of Porcelain Insulators on Power Transmission Lines based on Non-coaxial Infrared and Visible-light Images

As a critical component of power transmission lines, porcelain insulators are essential for the safety of China’s power grids. Traditional infrared imaging detection, limited by a constrained viewing angle and image resolution, often fails to detect local temperature rise problems. This paper aims to introduce a 3D reconstruction method based on visible-light-guided infrared images, utilizing the 3D Gaussian Splatting (3DGS) algorithm. The method generates denser point clouds from visible light images to serve as input for infrared reconstruction, and image registration is used to address the parameter differences between the two non-coaxial cameras. Experiments show that this method significantly improves the structural similarity index (SSIM) of the 3D reconstruction from 0.7348 to 0.9055 and the peak signal-to-noise ratio (PSNR) from 21.9887 to 23.1176 dB, enhancing the quality of infrared reconstruction of porcelain insulators. It can provide a practical foundation for realizing panoramic visualization of the thermal field on the insulator surface for accurate fault diagnosis.