Technologies
Recent software developments, commonly called ‘intelligent surveillance’ technologies
or ‘higher level automation’, will potentially allow the interpretation of imagery to:
• identify particular events, such as leaving an unattended bag or suspicious behaviour;
• recognise inappropriate motion, or movement in an inappropriate location;
• recognise car registration numbers and match them to a database; and
• recognise that an image of a person is very similar to a reference image.
Some technologies or systems currently available, and others under development, might
be applied to meet operational objectives in support of counter-terrorism in mass
passenger transport environments.
These systems, which deal with disparate data sources and data of differing quality, could
be developed to provide automated real-time and post-event system capabilities, such as:
• better capability to recognise suspicious behaviour, activity or objects and alert those
monitoring;
• better ability to identify individuals in a crowd, in poor lighting and when individuals
are partly obscured;
• better ability to track individuals in crowds and on ‘handover’ between systems or
cameras; and
• better CCTV data retrieval and analysis.
Limitations
The use of intelligent surveillance technology raises significant privacy and civil liberties
issues, and these must be balanced with its utility as a risk-based counter-terrorism and
law enforcement tool.
The technology is at a relatively immature stage of development and reliability for most
potential uses in mass passenger transport environments. It has been implemented with
only a limited range of simpler functions in larger scale systems. The scale and nature of
mass transit operations create practical difficulties for its continuous use. Difficulties
include inadequate imagery, wide areas to cover, the large number of cameras, the
potential for high false-alarm rates, and the relatively high cost.
Table B.1 Technologies for Real-time Situational Awareness
Application / Features/problems
requirement
Facial • Requires high frame rate and extensive database capability
recognition • Current face-in-a-crowd technology has low reliability
• Requires significant R&D and testing in real environment
Unattended • Current technology requires extensive configuration for individual
item environments, and can have high false-alarm rate
recognition • Further R&D required to reduce false-alarm rate and improve analysis of
video motion and events
Video motion • Varied sophistication of technology available – usually price based
detection • Can be unreliable in low light
• R&D required to improve performance in varying conditions and to
integrate into logic-based video analysis
External • Devices that detect environmental variations such as infra-red, pressure,
stimulus light/heat, smoke and/or respond to intruder alarm interfaces (intelligent
event and dumb)
triggering • Integrated in logic-based video analysis and subsequent event
management
• Further R&D would see ultimate solution based on video analysis, not
external components
Human– • Intelligent control room management, requiring only one person
machine Adaptable monitor configuration based on event status and video priority – all
interaction integrated into logic-based analysis system