Multiple Perspective Video Acquisition: Example

The following set of instructions describes how to acquire multiple streams of digital video data using equipment in the UCSD Visual Computing Laboratory.

  1. Find hammer (named for Mickey Spilane's fictional tough-guy New York private investigator) in the Visual Computing Laboratory. A picture of hammer, the computer, is below.

    This is hammer, the Windows NT workstation in the UCSD Visual Computing Laboratory.
  2. Make sure that the cameras that you want to use are set up around the scene that you want to survey and are connected to the video patch panel that sits beside hammer in the lab. A picture of the patch panel is below.
    This is the video patch panel that sit to the right of hammer.
  3. Check the video outputs from the cameras by patching them to the monitor input on the panel, one at a time, and verify that the desired signal is present. Patching means connecting the video signals on the panel with a short (about 1m) video (RG-59) cable. A picture of the camera video jacks is below.

    This picture shows a couple of the video jacks on the patch panel. There is a video and a synch line for each camera.
  4. For best performance, make sure that the video signals are synchronized. This means that all the cameras are refreshing their video at exactly the same time. To this by selecting one camera as a master and using its video as input to the synch of the other slave cameras. Remember not to split a video signal more than once to prevent degradation of the signal. If you need to split a signal more than that, use the video amplifiers shown in the image below. The yellow jacks are for the video signals. The top jack is the input and the bottom three are amplified video outputs. This acts as a one-into-three signal splitter with little degradation of the signal.

    This is what the video amplifiers on the patch panel look like.
  5. For this example, we will digitize from two cameras simultaneously at 30 frames per second, using one Matrox Meteor card per camera. The four video inputs for the Meteors are available on the patch panel. Connect one camera input to input 0 of Meteor 0 and the other to input 0 of Meteor 1. The Meteor inputs are shown in the image below.

    These are the Meteor input jacks on the video patch panel.
  6. Optional: Run Intellicam (Programs-Mil-Intellicam in the Windows NT Start menu) to verify that the Meteors are working. You may also want to use this to save a snap shop of the camera view for calibration at a later time. Exit Intellicam.
  7. Open an MS-DOS command shell (from the Windows NT Start menu).
  8. Change drive to E:.
  9. Change directory to \tmp.
  10. Run mpdvr to acquire video sequences. 
    
mpdvr -ncam 1 -seq0 0 -seq1 0 -n 600 ex0.raw ex1.raw
  11. Convert the raw files produced by mpdvr to MPIV0 format using mpiconvert. 
    
mpiconvert ex0.raw ex0.mpiv
mpiconvert ex1.raw ex1.mpiv
    Delete the raw files to save space.
  12. Optional: View the images by converting them to pnm (portable any map) format using mpivtopnm. 
    
mpivtopnm ex0.mpiv ex0
mpivtopnm ex1.mpiv ex1
  13. Transfer the MPIV0 files to am SGI workstation, preferably one with lots of space on its /tmp or /var/tmp partition.
  14. On the SGI workstation, make an SGI movie from the MPIV0 file.
    • create .ppm files from the MPIV0 file. 
      
mpivtopnm ex0.mpiv ex0
    • create the movie file 
      
makemovie -o ex0.rgb -f sgi -c jpeg -l once -r 30 -b ex0.????.ppm
    • view the movie 
      
movieplayer ex0.rgb
    • remove the .ppm files 
      
rm ex0.????.ppm

    •  repeat above for the other video sequences
  15. Optional: Make an MPEG movie from the MPIV0 file.

Web page design by Jeffrey E. Boyd