RAIL HEAD SURFACE DEFECT CLASSIFICATION USING ROUGH SET THEORY

摘要

Many machine vision based inspection systems applied to rail head surface defect detection for ensuring the safety of rail transportation during the past years.However, few of them focused on defect classification which is very essential for further identifying defect, and removing the undesired interferences, such as smudge, noise and rail gap.This paper completes rail head surface defect classification which includes two steps.First, a set of features is established using a fast feature extraction approach.Second, Rosetta system, which is a Rough Set Toolkit for data analysis,has been utilized to reveal classification rules.The effectiveness of feature extraction approach is verified, and a promising classification result is given.