Automated detection of inner surface defects in pipes using image processing algorithms

Abstract

This paper presents a new methodology for the automated inspection of pipes. Standard inspection systems are based on closed-circuit television cameras which are mounted on remotely controlled robots and connected to remote video recording devices. The main problems of such camera-based inspection systems are: 1) the lack of visibility in the interior of the pipes and 2) the poor quality of the obtained images because of difficult lighting conditions. The focus of this research is the automated detection and location of defects in the internal surface of pipes. The proposed optical system is an assembly of a CCD camera and a laser diode to create a ring-shaped pattern. The camera obtains images of the light projections on the pipe wall. A novel method for extracting and analyzing intensity variations in the obtained images is described. The image data analysis is based on image processing algorithms. Finally, an image of the pipe wall is generated by extracting the intensity information existing in the pipe pictures. Defects and anomalies can be detected using this extracted image.