Quick Look

Development of Automated Monitoring of Critical
Transportation Infrastructure, 10-9339

Printer Friendly Version

Principal Investigators
Michael J. Magee
Michael P. Rigney 

Inclusive Dates: 07/17/02 - 11/17/02

Background - Critical transportation infrastructure assets such as bridges, tunnels, and major corridors have long been recognized as vital components that contribute substantially to public safety and national economic activity. During periods of natural disasters such as hurricanes, for example, such assets are generally essential for providing the public with effective evacuation routes. From an economic standpoint, these assets provide vital links between major areas of manufacturing, supply, distribution, and consumption. The need to protect such assets is therefore of high priority within the context of homeland security.

A central issue with regard to protecting critical transportation infrastructure assets is the ability to monitor them effectively using closed circuit television (CCTV) cameras. The deployment of CCTV cameras has increased dramatically in recent years, but the number of cameras that individual operators can simultaneously monitor effectively is limited.

Approach - With the foregoing as background, the objective of this project was to develop and demonstrate proof-of-concept automated monitoring of critical transportation assets using imagery acquired from cameras viewing such assets. The primary emphasis of the project was to investigate, develop, and implement image-processing methodologies that can detect conditions that vary substantially from those that normally occur. In general, these methodologies consisted of the following four-step process:

  • An intensity characteristic model of the transportation asset was learned during a training phase.
  • During the anomaly detection phase, objects were segmented based on pixel characteristics that vary substantially from those embodied in the intensity characteristic model.
  • Segmented objects that matched certain morphological, topological, and/or geometric constraints were flagged as being candidates for further (temporal) processing.
  • Segmented objects with unanticipated temporal persistence or geometric characteristics were identified as being worthy of visual inspection by a human operator.

Accomplishments - Image sequences of traffic flow over a high volume overpass and the infrastructure supporting it were processed using the above approach. Specific test conditions were associated with different regions-of-interest such as the lanes of traffic and support structures under a freeway. Events associated with the different regions were flagged with informative messages indicating the nature of the event. The two illustrations show the nature of events that were successfully detected, identified, and associated with the specified regions. The first illustration represents a type of anomaly on the main surface of the roadway within which vehicles are normally in motion. In this case, a vehicle has stopped, with its position persisting for several seconds. Because this condition is not consistent with the normal state of traffic flow, it is flagged as anomalous. In the second illustration, an individual was detected moving under the freeway in an area where objects (such as people and vehicles) do not normally occur. This condition was appropriately flagged as an "Under Freeway Anomaly." 

The detection of both of these conditions is illustrated by activating the video

(a) Anomaly detected on main surface of roadway

(b) Zoomed-in anomaly

Persistent anomalous condition detected on main surface of roadway

(a) Anomaly detected under freeway

(b) Zoomed-in anomaly

Under Freeway Anomaly

2003 Program Home