Detecting possibly threatening conspicuous human behaviour in airport terminals via movement categorization

Abstract

Facing security issues is one of the most challenging aspects of global transportation. Detecting suspicious behaviour of passengers decreases the risk of an attack inside airport terminals. The preparation of attacks requires deception with concealment. In this paper we describe a methodology to detect categories of passenger movement which are based on analogies to visual and vocal signs of deception. We analyse position data for each passenger which is assumed to be derived by software examining raw surveillance camera data. Passengers are categorized by the shape of their path using physical quantities such as speed. Detected passengers are marked for further inspection. Testing and calibration of the proposed detector is done by a simulation environment covering a medium sized airport terminal building. After calibration, the detector is able to identify all previously defined conspicuous passengers. The approach can be applied to other traffic nodes.