New Scheme of Image Matching for The Power Transmission Line Icing

Line icing disaster on power transmission is a significant problem that causes severe problems in system security. Therefore, a reliable system is needed to monitor the Power Transmission Line Icing (PTLI). Nowadays, binocular cameras mounted on Unmanned Aerial vehicles (UAVs) have been widely used to observe cable conditions during icing. Existing three-dimensional measurement methods based on binocular stereo vision were recently introduced to measure the ice thickness in PTLI. An essential step in three-dimensional measurement is image matching. Image matching has been widely used with various methods, such as SIFT, SURF, and ORB. Compared with SIFT and SURF algorithms, ORB has absolute advantages in computational speed. However, conventional ORB still has some drawbacks when processing images that have a complex background. So this paper proposes a new efficient scheme to reduce errors and save time. Our approach achieves a 90% accuracy of the average distance between the two correspondence points in the left and right image. The standard deviation error value is smaller than conventional ORB, and the number of points matching errors is less than conventional ORB, even with almost no error matching on the tested images. The average processing time is ten times faster than conventional ORB. The simulation results show that the new scheme is effective for image matching on power transmission line icing.