Abstract: In view of problems of low efficiency, bad real-time performance and low accuracy existed in current detection algorithm of joint of powerful conveyor belt, a detection algorithm of joint twitching based on iterative threshold method for image segmentation was proposed. X-ray image is composed by data transmitted back from different detection plates, so iteration method is applied to the X-ray image for threshold segmentation firstly. Then according to characteristics of the joint, it is detected by Y-difference algorithm between the joint and background. Finally, the joints are matched based on distance between the joints, and vertical distance between the matched joints are calculated to realize detection of the joint twitching. The experimental result shows that the algorithm has high accuracy and rapid speed of detection, meets requirements of practical application, and provides guarantee for coal mine safety transportation.