The Digital Cherenkov Viewing Device (DCVD) is one instrument available to authority inspectors for verifying spent fuel assemblies in wet storage. The measurements result in images of the Cherenkov light emissions from the fuel assembly under study. This work presents research on applying image analysis and statistical methods to improve data quality and to extract more information from the measurements, extending the use of these methods beyond what is currently implemented in the DCVD software. The goal of this project is to apply template matching and statistical analysis to the images. However, before such techniques can be applied, effort is needed to ensure that the measurements are directly comparable. Two main issues are investigated here, the first being the positioning of the Region-Of-Interest. By developing an automated Region-Of-Interest placer, a consistent and reproducible Region-Of-Interest placement can be achieved. The second is automatic identification of fuel type, to support a later comparison with a template. We demonstrate that a method based on Principal Component Analysis can be used to determine the fuel type. Finally, we present the first results regarding template matching, comparing a measured image to a template, aiming to identify regions in the image where the two differ. Such differences could be due to a partial defect located in that region, but also due to other reasons such as debris covering the fuel top. Automatically identification of such regions can in the future be used to focus inspector attention to features requiring expert judgement, supporting efficient use of the measurement data and inspector effort. The first results demonstrate the feasibility of the method, but also that more work is required before the method is robust.
