3-D reconstruction of AGR fuel channels using RVI footage

Within the U.K, there is a fleet of 7 Advanced Gas-cooled Reactor (AGR) stations in operation which are close to or have already exceeded their original estimated design lifetime. As the reactors continue to age, it is critical to gain an understanding of the structural morphology of the AGR fuel channels and to quantify changes in the physical properties of the graphite bricks which occur during reactor operation. To assess such operational changes, periodic Remote Visual Inspection (RVI) is carried out on pre-selected channels using specialist inspection tools which gather video and other sensory data of the channel bore. Current visualisation of the video provides limited 3-D structural information due to the image acquisition process which results in a loss of depth information. This paper describes a novel image reconstruction framework based on a technique called Structure-from-Motion (SfM) which has been designed to ascertain 3-D structural information directly from 2-D in-core RVI data. This paper also discusses the challenges of applying 3-D reconstruction techniques like SfM to pre-existing inspection data before demonstrating that the proposed framework can be applied to data captured under laboratory conditions and to subsets of real in-core inspection videos to generate singular and circumferential point cloud reconstructions of the channel bore.