Vehicle license plate detection in images

Abstract

Frequently, there is a need to identify vehicle license plates (VLP) in images taken from a camera that is far away from the vehicle for security. The extracted information from vehicle license plates is used for enforcement, access-control, and flow management, e.g. to keep a time record for automatic payment calculations or to fight against crime. That's make license plates detection crucial and inevitable in the vehicle license plate recognition system. This paper aims to present a new robust method to detect and localize license plates in images. Especially we focus on the Moroccan's VLP. The proposed approach is based on edge features and characteristics of license plate's characters. Various images including Moroccan's VLP taken from different distances and under different angles were used to evaluate the proposed method. The experimental results show that our system can efficiently detect and localize the Moroccan's VLP in the images. Indeed, the recall/precision curve of the proposed method proves that 95% precision rate is obtained for recall rate value equals to 81%. In addition, the standard measure of quality is equal to 87.44 %.