Abstract: For problems of low sensitivity and complex detector with large volume of current longitudinal tearing detection methods of mine-used conveyor belts, a longitudinal tearing detection system of mine-used conveyor belt was designed by use of machine vision method. The system uses FAST corner detection algorithm and line detection algorithm based on Hough transform to detect longitudinal tearing phenomenon of conveyor belt, and adopts wireless sensor node for longitudinal tearing detection based on the proposed detection method and speed sensor to detect longitudinal tearing position and length of conveyor belt. The test result shows that the system can process 46.3 frames per second for a conveyor belt with length of 6 m, width of 800 mm and running speed of 5 m/s, and detection accuracy is 96.24%.