Institute of Communication Technics, Specialized High School (Cologne, Germany):

Thieling L., Prof., Dr. Eng., Scientific Fellow
Seegert S., Mag. Eng., Scientific Fellow, sebastian.seegert@fh-koeln.de

Embossed and stamped writing on metal surfaces is commonly used in the metal producing and processing industry. Products of modern continuous casting machines will be signed to ensure optimal quality control. Core approach of this paper is to consider the shadow cast on the labelled surface, the required specific lighting and in addition a concerted pre image processing algorithm. Subsequent image processing algorithms allow a position and dimension invariant  detection and scaling of the labelling. Scheme of the illumination and camera system for use in a casting plant. Afterwards classical OCR methods can be used for the generated results. Example of the intermediate results for the image pre-processing chain of the pattern „0940653“ is displayed and images of marking on the face of a round rod at about 1000 ^oC with horizontal and vertical illumination are shown. Results of the image pre-processing chain for a marking on a hot surface at about 1000 ^oC are anayzed.

1. Cai, J.; Zhang, G.; Zhou, Z.: Design of on-line automatic vision inspection system for steel billet characters, Chinese Journal of Sensors and Actors, Hangzhou, China, März 2006.
2. Jong-hak, L.; Sang-gug, P.; Soo-joong, K.: Vision technique for recognition of billet characters in the steel plant, 8th Pacific Rim Internat. Conf. on Artificial Inteligence, Auckland, Neuseeland, 9.–13. Aug. 2004, Springer Verlag. 
3. Seegert, S.: Prägeschrifterkennung mittels digitaler Bildverarbeitung, FH Köln, 2009.
4. Seegert, S.: Machbarkeitsanalyse zum Projekt „Einzelstabverfolgung“, FH Köln, 2010.