Image analysis methods for partial defect detection using tomographic images on nuclear fuel assemblies

A promising non-destructive assay method for verification of irradiated nuclear fuel is gammatomography, i.e. the use of measurements of the gamma radiation field around a nuclear fuel assembly to reconstruct detailed information about the internal source distribution.Typically, tomographic reconstructions result in two-dimensional images of cross sections of the fuel. We demonstrate how such images can be searched for fuel rods using a template matching technique, which is a method commonly used in the field of image analysis. In this case, a template or mask corresponding to the size and shape of a fuel rod is translated across the image in order to find the region with the highest reconstructed activity, which is assumed to correspond to the location of a fuel rod in the image. This is done iteratively, allowing no overlap of the rods. By defining the threshold between background and fuel rod objects in the image, we can identify and count the fuel rods using no other assumptions than the rod radius.Thus the rod identification procedure provides a possible means to verify whether all fuel rods arepresent, and it may also be implemented to identify the fuel type of the measured assembly. Theprocedure is robust in cases of irregularities, such as assembly bow or torsion, or the dislocation ofindividual fuel rods in the measured cross section.Here we demonstrate fuel rod identification procedure, using authentic images collected with a tomographic measurement device on commercial fuel assemblies. The results show that image analysis can support tomographic partial defect verification of irradiated nuclear fuel assemblies, even on the single fuel rod level.