VEHICLE AUTHENTICATION FROM DIGITAL IMAGE MEASUREMENTS

A. Psyllos, C. N. Anagnostopoulos, E. Kayafas

Abstract

In this work, it is demonstrated that relatively simple image processing functions based on symmetry measurements can be used to obtain high performance vehicle authentication, creating an “image signature” for every vehicle. On that basis, distinctive parts of the vehicle frontal view such as the headlights and the logo area can be segmented. A novel neural network application, properly trained, provides adequate vehicle manufacturer recognition when used in conjunction with a previously developed license plate recognition module. It is also shown that image processing functions based on Scale Invariant Feature Transform (SIFT) measurements can be used to obtain high performance object features location, creating a keypoint fingerprint for every feature. A grid of SIFT keypoints was drawn and a neural network was used also for vehicle model recognition.

Download

IMEKO-TC4-2008-009.pdf

DOI

-

IMEKO TC

TC4 - Measurement of Electrical Quantities

Back to the proceedings