Abstract: A recognition technology of two-dimensional code was proposed with an effective and accurate way for industrial applications, and the identification system for a batch of two-dimensional code with high speed based on machine vision was developed. Firstly, a method of locating the region of interest in each two-dimensional code with batch processing was proposed according to the geometric relationship between the measured two-dimensional code and the target subspace. Then, the anti-noise ability of the system was evaluated by adding Gaussian noise in the images, which simulated the noise produced in the actual process. Finally, the relationship between the recognition rate in the system and the motion speed of the two-dimensional code was analyzed, and the experimental results with references were compared. The results indicate that the average recognition time of each two-dimensional code is 17.8 ms with noiseless images and is 21.3 ms with noise images respectively in the test data set of 1 800 sheets with the motion speed of 296.8 mm/s, and the recognition rate is 100%. The system meets the online detection requirements.