Detecting Modifications in Printed Circuit Boards from Fuel Pump Controllers

Frauds involving illegal modifications to the printed circuit boards from fuel pump controllers are a serious problem, which not only harms customers, but also connects to other crimes, such as money laundering and tax evasion. The current state-of-practice for inspecting these boards is a visual analysis performed by a human. In this paper, we introduce an image-based approach that can provide support to the human inspector by automatically detecting suspicious regions in the boards. The proposed approach aligns a photograph of the inspected board to a reference view, partitions the image in sub-regions, extracts features using a variation of the popular Scale-Invariant Feature Transform, classifies the features against previously trained Support Vector Machines, and integrates the results for presentation. In experiments performed on a dataset containing 649 images from a board, with and without modifications, our approach achieved a precision of 0.7739, a recall of 0.9434, and an F-measure 0.8503. These results indicate that our approach can effectively identify suspicious regions, providing invaluable help to the human inspector.