Research on Character Recognition Technology for Curved Surface in Industrial Application Environment

The material tracking technology in the manufacturing industry is the cornerstone of intelligent manufacturing, and the material tracking technology needs the timely and effective recognition of the identification for characters printed on the surface of the material as a support. In this paper, a recognition technique for characters on curved surface of materials in industrial environment is proposed to solve the shortcomings of conventional scene text recognition methods, such as processing deformed characters on curved surfaces and the situation of insufficient training samples. Firstly, the image is transformed into a binary graph suitable for detection algorithm by adaptive hybrid threshold binarization algorithm. Then, the character location algorithm based on DBNet semantic segmentation model and its corresponding post-processing algorithm are used to locate and correct the text line region, and the recognition is completed by single character split and recognition. Finally, the method proposed in this paper is compared and verified. The results show that the targeted improvement for each step of curved surface character recognition in this paper has achieved certain improvement in recognition accuracy index and visual effect, and has certain engineering application feasibility.