Design of label information recognition system based on vision

With the frequent use of label form in daily life, it is difficult for human eyes to observe the label information efficiently and precisely to meet the requirements of modern industrial production. Aiming at the above problem, the theory research and system design of label identification system have been carried out. In this paper, the proposed label recognition technology is mainly based on the core idea of creating template tags and the main technology of image processing as the premise. Using the relative position relationship between template tags and the target area for positioning, it can process the identification of tags entering the acquisition area at any state or from any angle. Subsequently, image preprocessing, two-dimensional affine change and morphological processing of the image were performed to make the image clear and concise to the greatest extent. On this basis, from the location of the text to segment the interested region and through the template matching method to identify the single text, then output the final result as a string. Finally, using the designed interface to visualize the application effect of the whole system. The advantages of the whole design lie in: first, it only needs to be initialized once, then it can enter the loop to identify the designated position, which is highly automated; the second is that the system can identify the label form of different angles and states, which makes the recognition range of the system wider.