Features

• Developed as a counter-terrorist technology to augment security, surveillance, and safeguards functions
• Detects motion in an area of interest in the camera field of view, while simultaneously filtering out nuisance motions
• Operates solely by processing video images - there are no other sensors or detectors involved
• Able to rapidly reconstruct events in true 3D space by combining information from three or more cameras
• Determines the size and shape of detected objects
• Permits activities of interest to be documented as they occur and provides a mechanism for triggering alarms
• Smart Video applications include:
  • Nuclear facilities such as reactors, fuel processing, and fuel and waste storage areas
  • Weapons storage and dismantlement areas
  • Restricted areas of airports

Description

Smart Video was developed through a collaborative effort between CANBERRA and the General Physics Institute (GPI) of the Russian Academy of Sciences - a joint U.S.-Russian effort sponsored by Brookhaven National Laboratory (BNL) through the IPP Program - to provide a more effective and reliable solution for safeguards systems. Smart Video integrates the safeguards and network computing capabilities of CANBERRA with the physics and analysis capabilities of the GPI. Smart Video was developed as a counter-terrorist technology to augment security, surveillance, and safeguards functions in nuclear facilities where terrorist actions could have global consequences of disastrous proportions. In its safeguards role, Smart Video operates totally autonomously, recording its observations for later review and analysis.

Smart Video allows an untiring computer to augment human security personnel as they monitor video projected from areas under surveillance. The Smart Video system automatically detects and traces the movement of people and materials and generates alarms if specified activities are performed. The alerts and alarms are computer signals that might be used to place a phone call to a SWAT team, to close and lock doors and gates, to activate sirens and flashing lights, or to generate any other response that might be appropriate to the specific installation.

The computer automatically generates an alert signal whenever it recognizes characteristic shapes moving around in the area of interest. It also automatically generates an alarm whenever an object moves into a defined sensitive area. For example, anyone approaching a control panel or the door to a nuclear material storage vault would trigger an alarm. An alarm is also generated when the Smart Video system detects that something has been removed from or added to a defined area. Therefore, theft of a sensitive item or depositing an object, such as a briefcase or backpack in a sensitive area, would trigger an alarm.

Example applications of Smart Video technology include:

• Nuclear facilities such as reactors, fuel processing, and fuel and waste storage areas
• Weapons storage and dismantlement areas
• CBW storage and dismantlement areas
• Restricted areas of airports
• Detection of unattended baggage in airports
• Arms Control Treaty Verification
• Monitoring evidence storage facilities
• Warehouse monitoring

THEORY OF OPERATION

Smart Video brings a new dimension to computer assisted video surveillance - literally. By combining information from three or more cameras, Smart Video is able to very rapidly reconstruct events in true 3D space. The 3D processing completely avoids, rather than solves, the problems of inherent ambiguities in interpreting 3D events on the 2D camera detector plane. This dimensional extension allows objects in motion to be detected, located, categorized, and often identified on the basis of their size, shape, and movement in real space. An object's true dimensions can be determined anywhere in the volume covered by the cameras. While phenomena due to lighting changes such as shadows, reflections, and glint are interpreted as motion events in 2D-space, these phenomena have no substance, and have no volume; and so are discarded in 3D space. Consequently, false alarms from these effects are eliminated.

The intent of Smart Video is that, without human direction or intervention, the camera system will detect movement of objects, identify any "objects of interest", and record their trajectory. Smart Video can dramatically reduce the volume of data collected in an unattended or remote monitoring scenario by filtering out motion alarms that do not involve specific objects of interest. Smart Video's ability to recognize specific objects and record their movements will have immediately beneficial application to safeguards of activities such as nuclear fuel transfers, materials movements, refueling, and other similar activities that are so complex that they currently either require an inspector to be physically present, or the monitoring systems generate huge amounts of data, much of it redundant, that must be reviewed by the inspector.

Because Smart Video automatically detects and traces the movement of people and materials, these events can be brought to the attention of the security personnel. Moving objects can be highlighted by a marker on the display screen for the guard to evaluate. Objects that are deposited or removed from the area can be highlighted for review. The 3D capability allows movements toward control panels or doorways to be highlighted. The highlighting could be augmented to include computer signals that automatically place a phone call to a SWAT team, close and lock doors and gates, activate sirens and flashing lights, or generate any other response that might be appropriate to the specific installation.

Smart Video operates solely by processing video images - there are no other sensors or detectors involved. Five times a second, each camera processor captures an image and processes it to detect any differences from the reference image.

If differences are detected, the image is further processed to eliminate noise and isolate groupings of pixels that represent the changes (as many as several hundred groupings can be present). The physical parameters of the groupings are computed (e.g. centroid, extents, moments of inertia, etc.) and transmitted to the central server. At the central server, camera placement information is combined with the grouping parameters from all of the cameras to construct the sequence of intersecting planes that represents the surface that encloses the 3D volume of each grouping. The centroid and moments of each enclosed volume are then computed. The absolute position of each enclosed object (x, y, z) is determined from its centroid, and the size and shape of each object is determined from its moments.

The Smart Video camera processor can either be mounted in the camera enclosure itself or in a central control room. In general, existing camera installations can be augmented as-is, provided there are a sufficient number of cameras, and their fields of view provide sufficient coverage of the area of interest. The central server is an off-the-shelf high-end computer. For situations in which the video itself must be used as legal evidence, the camera processor must be mounted in a tamper-proof enclosure with the video camera. Each picture is cryptographically authenticated with an RSA signature that proves that the picture is unaltered.