ABSTRACT

In recent years, ATM attacks have grown owing to flaws in the existing security systems. For this reason, many ATM premises have video surveillance systems installed to monitor crime. Nevertheless, it is extremely difficult for security officers in 24×7 days monitor in real-time. After an incident, security officers must manually search the incident videos from a huge collection, and the system will never detect and warn of criminal activities. This paper proposes the algorithms based on mathematical model for intelligent video surveillance systems to prevent the occurrence of criminal activities at ATM premises. There are two algorithms developed that address the two most critical issues facing ATM security services: ATM cash robbery detection and machine steal detection are based on frame differences and contour-based Approaches. Proposed algorithms are proven to provide accurate results with less time and less computational cost through excellent performance.

KEYWORDS: ATM, Cash Robbery, Machine Steal, Contour Area, Motion Detection, Object Tracking. Shape Analysis.