Abstract: In this paper, according to the characteristics of basic feature and identification about onboard equipment DMI cab signals of CTCS-3 level Train Control System, a novel method based on Cluster Analysis and SVM Algorithm was proposed for DMI cab signal automatic identification. The paper established a model of cab signal, extracted the target features by binary image processing techniques, classified cab signal image preliminary cluster analysis made target identification for various cab signals by SVM classifier, provided a method supporting for further automatic data collection of DMI cab signals. Results showed that the novel method was with good performance. The average image recognition rate was reached about 95%.