DIGITAL IMAGE ANALYSIS DEFECT INSPECTION TECHNIQUES USING THE FLUORESCENT DYE PENETRANT INSPECTION PROCESS

摘要

This paper describes the development of a comprehensive set of image analysis tools, which can be used to produce an automated inspection system. The aim of the image analysis is to detect the presence of surface breaking material defects such as porosity, cracking and machining defects. A non destructive testing (NDT) method known as fluorescent dye penetrant inspection (FPI) can be used to highlight the presence of defects, allowing defects as small as small as 0.5 microns wide by 5 microns deep to be detected. Often complex geometry components must be imaged this has led to the development of flexible software tools which can be quickly tailored to a particular inspection application. This paper gives a brief introduction to the NDT method of FPI, the aims and objectives and motivation for this work is then described. A brief review of other research in the same field is also presented, this leads onto the main section of this paper which details the image analysis tools developed namely; image pre-processing, interactive region of interest selection/auto region selection, Image calibration, image labelling/recording skilled inspectors knowledge, image classification (intelligent decision system), decision rule optimisation, This paper also focuses upon the results obtained from the system and the performance monitoring methods used. Finally conclusions are drawn and future work is presented. The developments detailed in this report are part of a project to improve and standardise the fluorescent penetrant inspection (FPI) process.