Text and character recognition on metal-sheets

To improve the automation of metal sheet production these sheets have to be tracked during the processing steps. This is preferably done by video optical tracking. To recognize the metal sheets each of them has an unique ID imprinted. This paper describes a method to automatically detect the characters on metal sheets, to classify them and to combine them to a string. First the image is preprocessed to detect the metal sheets in the image. After this step every character is located and classified. Finally all characters are combined to one string with its specific certainty. The algorithm has to determine about the alignment and the orientation of the text because it is not known before. The most problematic part is the extraction of the objects, which may be characters, due to difficult lightning situation. There can be different lightning spots visible in the image because of the movement of the sun, weather and factory infrastructure (windows, ceiling lighting) which cause sometimes difficult reflections on the metal sheets. The introduced program is completely written in Matlab 2015b. It has been applied to a wide range of images with successful results.