Abstract: In China,the automation level in some steel processing plants is low,which the weld was manually judged and the position of the welding torch was manually adjusted,and the welding quality is influenced by visual fatigue caused due to long-term work. The image information was captured through the OpenM V module,real-time weld information was extracted by using related image processing technology,the offset of the torch was determined through data statistics,and the automatic correction of the torch was realized by combining control technology.The field test results show that the system can achieve automatic weld tracking with high stability and accuracy.