Automatic Detection of Pipeline Components in Radiographs for Corrosion Monitoring

Abstract

This paper introduces new techniques to support industrial radiographic inspection, aiming at automatic corrosion monitoring in pipeline systems. Using the methodologies we proposed, pipeline components to be inspected are automatically detected in the radiographic image, their wall thicknesses are measured and parameters for corrosion detection are computed. For the automatic detection and recognition of pipeline components, we developed a new method that uses image matching techniques in conjunction with genetic algorithms. For measuring the pipe wall thicknesses we used image segmentation techniques based on the analysis of image line profiles. As a result, analysis of pipeline radiographs for corrosion monitoring can be automatically performed, improving the reliability and speed of the inspection process.