Automatic Vehicle Detection and Driver Identification Framework for Secure Vehicle Parking

Abstract

In recent times, automatic face recognition algorithms are playing a key role in several security applications. In this paper, we develop a framework for enhancing the security of vehicle parking spaces. The proposed framework can be divided in to three separate steps. In first step, a vehicle in the input image is spotted. In second step, driver face is located. In final step, a robust face recognition algorithm identifies the driver by comparing the face image with face images in a database. On successful identification of the driver face, vehicle is allowed to enter in parking area. To detect vehicle and face(s), we use Adaptive Boosting algorithm and Haar-like features, while driver face identification algorithm uses Eigenfaces for feature selection and Euclidian distance for classification. To test the face identification, we simulate a challenging situation where only a single facial image of a driver is available in the database and four face images in different poses are used for testing. Simulation results show very high detection and identification results regardless of the facial pose variation. The results demonstrate the feasibility of developed framework to be deployed in any public vehicle parking area.