Automatic detection of manipulated packages by image comparison
Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesis
Shipping companies have a lot of difficulties to control and screen packages such as mail or different types of cargo that may contain dangerous or illegal goods. This project has the goal to enhance the supply chain security by elaborating a way to automatically detect manipulated packages. Based on image comparison, an algorithm can say whether a package shows significant doubts to be put aside for more inspection. After being coded, this algorithm is explained in this thesis. It uses an image registration method as well as a differentiation technique to detect tampered patterns from one image of package at one point of the supply chain to another image of the same package at a later stage. Results show that re-sealed packages, ripped packages or blacked labels can be detected and minor manipulations such as modifications on the barcode can be detected with a close-up of the local region. The principal goal of this project is to give recommendations on what is possible to detect and how to implement this algorithm. This work is a first step toward the use of image processing in future work for package screening.
Place, publisher, year, edition, pages
intensity-based feature-based registration differentiation comparison suspicious manipulated packages
IdentifiersURN: urn:nbn:se:kth:diva-182103OAI: oai:DiVA.org:kth-182103DiVA: diva2:903259
2015-11-11, 11:55 (English)