Automated image stitching for fuel channel inspection of AGR cores

Visual inspection of fuel channels is an important element of the understanding of the health of the current fleet of AGR reactors. When a fuel channel is inspected, video footage of the entire inside surface is recorded through a series of vertical scans of the channel. When areas of interest such as cracks are identified, screenshots of these areas are taken and manually stitched together to produce a montage of the region of interest. This is a lengthy process, which requires an experienced person to undertake. The resultant montages are assessed and then included in the TV GAP sheet, a document that forms part of the case for return to service. This paper describes an automated approach which uses advanced image processing techniques to recreate a full 360° image of the inside surface of the channel using the same video input. These images offer a significant improvement in the quality over the manual approach, provides 100% coverage of the channel and can be generated in a fraction of the time of the manual images. The software has been applied to over 30 recent channel inspections, and has been demonstrated using footage from all 7 AGR stations.