MPI-Video Surveillance and Monitoring:
Test Bed

As part of our efforts in video surveillance and monitoring at the Visual Computing Laboratory of UCSD, we have set up a test bed for developing MPI-Video systems. The photograph below shows a stylized view of the test bed.

A stylized view of the UCSD Visual Computing Laboratory's MPI-Video test bed.

The room on the right-hand side of the figure is the lab's reading room and library. We use it as an area for surveillance because it is convenient. It is next door to the lab (on the left-hand side of the figure). The figures below show two more stylized views of each of the two rooms.

A stylized view of the library/reading room that is part of the MPI-Video test bed.

A stylized view of the laboratory portion of the MPI-Video test bed.

We have placed a set of cameras to view activity in the library. The two photographs below show a view of the library with a camera mounted on top of a book case, and a close-up view of one of the cameras.

A view of the library under surveillance. One of the video cameras is visible, mounted on top of the book case.

One of our Sony XC-999 surveillance cameras installed in library.

Coaxial cables (shown as black lines in the first figure) carry the analog video signals from each of the cameras to the video distribution center in the laboratory. This distribution center, shown below, is a 19" rack with a set of panels with BNC bulkhead through connectors mounted. All of the video signals from the cameras are broken out to the front of the panels via the through connectors. We also bring all of the video input lines from each of the video workstations (described later) to the panel. This allows us to quickly route any camera's video signal to any computer. Changing the distribution is easy allowing researchers ease and flexibility in conducting experiments. For outdoor experiments, we bring video signals from exterior cameras to the panel via RF links and close-circuit TV. A monitor sits on top of the 19" rack with its input connected to the panel so that researchers can quickly verify a video signal. Also, there are two low-cost audio-video distribution amplifiers mounted in the cabinet to allow branching without loss of signal.

The video distribution center.

We have a variety of video workstations. These include:

A Pentium (233MHz) with two Matrox Meteor video acquisition cards (aka Hammer). This machine is the workhorse of MPI-Video data acquisition. It can capture two color video signals simultaneously, six black and white, or can multiplex through as many as eight signals.
Two Pentiums (166MHz) with Miro DC-30s. These machines can capture only a single channel of video and have not multiplexing capability.
Two SGI O2s with video boards (aka Leaphorn and Chee). Like the Miro DC-30s, these machines can only capture from one video channel and cannot multiplex.
A collection of SGI Indys and Indigos.

Two of the workstations, Hammer and Leaphorn, are shown in the figure below, sitting directly to the left of the video distribution center.

Two video workstations, Leaporn (left) and Hammer (right).

On feature of our facility that makes our test bed flexible and expandable is the use of PVC conduit (we used a generic version of the well-known Panduit). All of the video signals are routed through this conduit. The covers of the conduit are easily removed to allow researches to reroute cables as needed. Once the cables are routed, the covers are placed back over the conduit, keeping cables out of the way. The photographs below show the conduit open, exposing the cables inside, and the covered.

Conduit with cover removed to expose cables within.

Conduit with cover in place.

MPI VSAM Home Page -- Executive Summary -- Our Vision -- Project Information --
Papers -- Investigators -- Administrative Staff -- Related Links

Web page design by Jeffrey E. Boyd

MPI-Video Surveillance and Monitoring:Test Bed

MPI-Video Surveillance and Monitoring:
Test Bed