Applying image analysis techniques to tomographic images of irradiated nuclear fuel assemblies

In this paper we present a set of image analysis techniques used for extraction of information from cross-sectional images of nuclear fuel assemblies, achieved from gamma emission tomography measurements. These techniques are based on template matching, an established method for identifying objects with known properties in images. We demonstrate a rod template matching algorithm for identification and counting of the fuel rods present in the image. This technique may be applicable in nuclear safeguards inspections, because of the potential of verifying the presence of all fuel rods, or potentially discovering any that are missing. We also demonstrate the accurate determination of the position of a fuel assembly, or parts of the assembly, within the imaged area. Accurate knowledge of the assembly position enables detailed modelling of the gamma transport through the fuel, which in turn is needed to make tomographic reconstructions quantifying the activity in each fuel rod with high precision. Using the full gamma energy spectrum, details about the location of different gamma-emitting isotopes within the fuel assembly can be extracted. We also demonstrate the capability to determine the position of supporting parts of the nuclear fuel assembly through their attenuating effect on the gamma rays emitted from the fuel. Altogether this enhances the capabilities of non-destructive nuclear fuel characterization.