Abstract: There are certain deficiencies in detection methods on the traditional electric locomotive contact wire geometry parameters. Based on the testing requirements and image characteristics of the worn wire, the integrated measurement method of online, non-contact, multi-parameter was proposed based on binocular vision technology. In-depth theoretical analysis had been conducted on image edge detection based on LOG algorithms, on which new methods of edge location by means of energy center algorithm were proposed, and numerical simulations were conducted. Simulation results show that the algorithm of energy centers has got stronger anti-jamming and anti-blur capabilities, with higher positioning accuracy. Verification experiments are carried out by using round steel with different heights, stagger values and wear widths as the detection object, and it is compared with the results tested by the laser rangefinders and digital display vernier calipers. Experimental results show that the edge detection by means of energy center algorithm is more accurater than LOG algorithms. Simultaneously, the measurement method proposed in this paper has smaller indication errors, better reproducibility, and it can meet the integrated needs of online, non-contact testing on the contact line.