Automation of microstructure analysis by artificial intelligence image processing

The paper presents the basic principles of automated microstructure analysis by combined Artificial Intelligence tools, such as empirical rules, fuzzy logic and neural networks: Kohonen LVQ (Linear Vector Quantifier), probabilistic, fuzzy ART (Adaptive Resonance Theory) and back propagation. It is described the methodology for ''intelligent'' filtering of relevant information about material degradation stage from metallurgical microstructure images taken by replica technique. Machine learning methods combined with advanced image modelling are applied for development of neural network correlations for automated damage pattern recognition.