In the iron front system, a large number of belts are used for rapid material transportation. The belts operate in the harsh environment of high load for a long time, and often cause belt surface defects due to large foreign body stuck, wear and other reasons; or the materials transported are at high temperature, which will also cause belt surface cracking and other damages. These minor damages are not noticeable and not easy to be found, but after the defects are expanded, the belt is prone to tear accidents, which will cause long-term unplanned shutdown and major economic losses. Because some belts are located at a higher position or are difficult to reach, especially the high-altitude conveyor belt, the personnel can not completely cover the spot inspection at ordinary times, and can not grasp the defects of the belt in real time; once the unplanned shutdown is produced, the shutdown and losses caused are often difficult to evaluate and make up for. In order to effectively avoid this situation, we began to focus on the avoidance of belt defects and the research and development of intelligent testing. According to machine vision, when there are small damages and cracks on the belt surface, the corresponding alarm information will be given, and the location of belt damage will be accurately reported, so that the relevant maintenance personnel can do their work in advance, reduce the shutdown and loss caused by unplanned maintenance, which is a good solution for belt system management. While the site for a long time purely rely on manual inspection of the belt to find the surface defects of the belt, there are often many dead corners, and can not be real-time defects of the belt to the maintenance personnel feedback, so that both increased labor costs and reduced the efficiency of the enterprise.