Study Of Rolling Mill Production Line Of Heavy Rail Steel Billet Recognition System

OCR, which refers to Optical Character Recognition, is an important part of artificial intelligence and pattern recognition. It has been practically applied to many traditional fields, this paper applies this kind of new technology to steel industry. In the actual situation of heavy rail production line of rolling mill, those factors like lighting, temperature etc. make it difficult to use OCR. Aiming at those problems, this paper advances some feasible solutions for the purpose of improving intelligent production level of steel.In the context of recognition system of billet, this paper systematically summarizes previous research results and further analysis the new problems that occur in practical application. By a larger number of experiments and practical application, this paper verifies the feasibility and reliability of character positioning of billet. It has basically improved recognition rate of the recognition system and provided reference for other application fields.Firstly, the billet character image preprocessing algorithm is studied, including grayscale image filtering and image enhancement, different methods are analyzed for the billet image processing effects. Secondly, the segmentation of billet character image has made a lot of analysis and comparison, and designed a recursive Otsu segmentation criterion for multi-level filtering and combined the segmentation and positioning of the characters. A character of billet positioning method based on projective invariants is proposed, using the characteristics of the target character and the definition of projective invariants measurements. The region of target character has been quickly found according to successive segmentations billet character images, and the target location information of characters has been outputted in the end. Then a multistage projection split algorithm is proposed for the areas of adhesion character, those have laid a good foundation for the subsequent character recognition. Finally, the recognition interface for the software system is properly improved in order to enhance the human-computer interaction.Experimental results show that this system can successfully solve problems of billet character recognition that occur on heavy rail of rolling mill production line. The recognition rate of the characters is 95%. It has certain suitability and stability in complex situation of production line, and at the same time, provides a kind of efficient solution to the system.