Automatic Rotation of the Copper-Brass Seal's Images During Their Verification

Abstract

International Atomic Energy Agency (IAEA) and European Atomic Energy Community (EURATOM) use thousands of Copper-Brass Seals every year to keep continuity of knowledge of nuclear material. The verification, done at headquarters is a time consuming procedure, whose operations are performed almost manually, i.e. the seals under verification are manually rotated until they can be compared to the image stored in the Database (DB) that was taken before sending the seal on the field. We are proposing an algorithm for automatic rotation of the image to match with the reference image from the DB. The main considered parameters are reliability of the algorithm and speed. A set of couple seal images was captured by an industrial vision system and pre-processed. Programming code for image rotation in each degree was developed, which estimates the similarity between the original and rotated image by the following parameters: Euclidean Distance (ED) and Sum of Absolute Differences (SAD) and returns the angle, for which one the min value of ED and SAD were found. The program was tested for 304 digital images of real seals, provided from IAEA and EURATOM. The maximum error during the experimental determination of angle of seal rotation was 0.3%. A proposal for a faster algorithm for rotation and determination of the rotation angle of the image related to the reference image is made. The proposed algorithm was tested for the same set of images