Abstract: Thermal cameras have widespread applications in smart grid layout optimization, equipment monitoring, and fault diagnosis. The fusion technology of thermal images and point clouds has attracted attention in the power industry. However, effectively integrating it with existing systems remains a critical issue that needs to be addressed. To efficiently fuse and display a single thermal image and point cloud while ensuring flexible server deployment and decoupling from existing system functions, this paper proposes a high-precision infrared and point cloud registration applied in a microservice-based digital twin system. The system performs calculations and optimizes the data update strategy on the server side. First, the matching feature points are calculated using the thermal and point cloud registration algorithm, followed by determining the three-dimensional coordinates of the thermal image feature points within the point cloud scene. Subsequently, the PnP algorithm is utilized to calculate the pose of the infrared camera, ensuring precise alignment of the thermal image with the point cloud scene. Finally, projection information is calculated using the rasterization algorithm, and the thermal values from the the thermal image are mapped onto the three-dimensional point cloud, enabling efficient fusion. The system is employed in infrared monitoring systems for power equipment. Real large-scale outdoor experiments have demonstrated that the accuracy of feature matching exceeds 96.17%, and the average projection error of correct matching is less than 7 pixels.