Abstract: The train number is a unique identification of the train. The Trouble of moving EMU Detection System (TEDS) is aimed to detect the trouble of moving EMU images based on the matching between present image and previous one which searches in the image database of the train number. However the RFID electronic tag does not installed in the EMU. Hence, the EMU train number recognition has become urgent to solve the problem. The article proposed a template matching algorithm based on semantic symbiosis probability to identify the train number. The experimental results showed that the correct rate and effectiveness of the proposed approach could conform with the regulation of China Railways, ensure the effect of TEDS.