Recognition of distorted characters printed on metal using fuzzy logic methods

This paper is based on real application whose task was to recognize characters printed on metal ingots. The problem is that the surface of ingots is very uneven - ingots are either hot, or cut by rough instrument, the printing machine can be worn down, etc. Therefore, the first difficulty was to separate characters from the background and the second one was the fact that the separated characters are heavily distorted. It is a challenge to recognize the letters correctly. In this paper, we will present two approaches: recognition based on application of mathematical fuzzy logic and a special human behavior inspired approach.