The paper aims to propose a marking area positioning method based on machine vision to improve the effi-ciency and positioning accuracy of marking, and reduce the labor intensity. Firstly, a template database was established to obtain the surface image of the cans with different patterns to be marked, and then the image was back projected on the cylinder surface to expand the image with a certain radian, and the image was spliced by the template matching principle to get the complete image of the can surface; secondly, ORB was used to determine the types of the cans surface pattern in the template database, and the template matching method was used to determine the position of the marking area of the type pattern on the can to be marked. The experimental results showed that the method can locate the marking area of 40-65 mm can quickly, and the error range was ±2°. Marking area positioning method for cans based on machine vision can realize the automation of the marking process, locate the marking area quickly and accurately. It has high flexibility, reduces the labor cost, and meets the production and processing needs of enterprises.