A proposed framework for automated real-time detection of suspicious vehicles

Abstract

Automated license plate recognition system only is not enough to detect suspicious passing vehicles. In this paper, we present a proposed framework for automated real-time detection of suspicious vehicles. It consists of two sub-systems, an automated license plate recognition system and a vehicle model recognition system. An automated license plate recognition system consists of license plate detection, character segmentation, and character recognition. A vehicle model recognition system consists of vehicle segmentation, license plate localization, taillight detection, vehicle class identification, model recognition. Finally, the results obtained from two systems are compared with the data obtained from the department of land transport in real-time. The proposed system is able to monitor suspicious vehicles and generates the reports or suspicious transactions to the department of land transport and third party.