Fuzzy logic system for superficial defects classification in metal patches
The system achieved is an integration of the identification system, we planned and carried out, as we previously described (Niola et al., 1999). From the depth map onwards, that is the output of the identification system, the classification system, operating on the single lines of the map, reconstructs through interpolation the level lines of the image, which are to be fitted to Fuzzy set; in order to do this, a mixed neuro-fuzzy technique is adopted. The so achieved input Fuzzy sets, level lines that have been modelled, are compared to the output Fuzzy sets, which reproduce a parameterisation of the defect according to its depth, by means of inference rules planned on purpose. The final result will give a quantification of the defect found, besides pointing out the area in which it has been noticed.
Bibliogr. 5 poz., wykr.