Abstract: With the extensive application of UAV technology in the field of power line inspection, how to efficiently and accurately process the multispectral images captured by UAVs has become a significant issue. This paper addresses the field-of-view difference issue in the spectral images of transmission line insulators captured by UAV-mounted multispectral cameras during the patrolling process by proposing an intelligent alignment algorithm for non-coaxial spectral images based on YOLO-SuperPoint. The algorithm first uses the YOLOv5 model for precise localization of insulators and removal of background interference, then extracts image features through the SuperPoint algorithm, and finally evaluates performance using integrated mutual information as the alignment index. Experimental results demonstrate that the algorithm can effectively achieve high-precision image alignment, with an average integrated mutual information value exceeding 0.8, significantly outperforming traditional alignment methods. This study provides a new technical means for image processing in UAV inspections, which is conducive to enhancing the level of intelligent power line inspection.