Abstract: Transmission lines often develop hidden fire sources due to internal equipment failures or vegetation obstruction, making direct visual observation difficult and reducing the accuracy of fire point localization. Infrared imaging can detect temperature variations caused by fires, enabling early identification of potential ignition points even before open flames become visible in visible-light images. Therefore, this paper proposes an accurate infrared positioning method for fire points in transmission line video surveillance images. Based on infrared imaging principles, the method acquires infrared surveillance images and preprocesses the raw images through filtering and enhancement. Image features are extracted and compared with reference features of fire points to determine the pixel coordinates of the ignition points in the monitored images. Using a three-dimensional Cartesian coordinate mapping relationship, the precise spatial coordinates of the fire points are calculated, thereby completing infrared positioning. Performance tests demonstrate that, compared with conventional localization methods, the proposed approach reduces the average positioning error by approximately 14 and 17 m in DC and AC transmission line scenarios, respectively.