Automated Pipe Inspection Based on Image Processing

The gas that keeps us warm or water that we need for our survival are all transported to our crowded cities from rural areas via many underground pipelines. Generally, the inspection of these pipelines is performed visually using videos recorded by a camera mounted on a mobile system which moves inside the pipelines and controlled by an operator. The video recorded by this mobile system is later analysed by operators to find anomalies that may present at the interior of these pipelines. However, this method is human dependent and rather slow. This paper introduces a highly automated system that is capable of performing fast and high precision inspection of dents via creating 3D profiles using a special laser profiler mechanism and image processing algorithms. In the proposed system, a laser source is used to project a laser beam onto a canonical mirror to form a ring shape laser light on the pipe's internal surface Images of the laser ring profile are acquired with an omnidirectional camera; the circular shape of the laser ring profile disturbed when the laser passes through dents. A fast boundary following is developed to detect these changes in the profile by computing Euclidean distance between each measured point on the laser ring and the ideal centre of the profile. The locations and characteristics of dents are determined by means of odometry sensors and the proposed image processing techniques, respectively.