Rust detection in images of metal alloys using Paraconsistent Analysis Networks (PAN) João Luís Lopes Freitas, Aldo Ramos Santos

Abstract

This work describes the construction of a  PAN  capable of performing the binary
classification of images of  metal alloys  for the presence of rust. The  considered  input parameters follow the texture characteristics studied by Haralick  and were able to  produce  a 70% accuracy  response  rate without  the execution of  the steps of parameters  selection  and
tuning.