Improvement of intelligent framework for suspect vehicle detection system

The crime problems become critical issues for national security especially the security of border and intelligent transportation systems (ITSs). As a result, the automatic suspect vehicle detection emerges as one of effective tools to tackle the problems. The law enforcement information and data generated by vehicle sensors are necessary to identify criminal or suspect vehicles that crossing the checkpoint. However, the traditional process is not effective and accurate. It is unable to classify the criminal or suspect vehicles if they are forged license plate or changed color, brand or type illegally. This paper proposes a new framework for automatic suspect-vehicle detection, with an improvement of illegal vehicle classification. It compares legal vehicle records obtained from Department of Land Transport (DLT) with those of vehicles crossing checkpoint. Also, DLT's legal vehicle records can be used to improve criminal or suspect vehicle detection. From extensive experiments, the results show that the proposed framework can increase the detection accuracy rate 27.51% beyond the traditional counterpart. In addition, the new system can detect illegal vehicles, criminal vehicles, and suspect vehicles although they are seen with forged license plates.