Perimeter intrusion prediction method using trajectory frequency and Naive Bayes classifier

Abstract

This paper proposes a novel perimeter intrusion prediction algorithm that can be applied to generic perimeter security systems. The proposed algorithm uses multiple probability mass functions that are computed using trajectory frequency information for different behaviour models: non-intrusive intention and intrusive intention. A Naive Bayes classifier is used to compute the intention probability for integrating multiple probabilities from different probability mass functions. The performance of the proposed algorithm is validated by numerical simulations and the classification characteristics are also discussed.