Re: Mars 3-D

[P3D John W Roberts <roberts@xxxxxxxxxxxxxxxxx>]

>Date: Fri, 11 Jul 1997 09:14:56 -0500
>From: P3D Sandman  <sandler@xxxxxxxxxxxxxx>
>Subject: Re: Mars 3-D 

>P3D John W Roberts says:
>>The interocular on both the lander and the rover are not too far off from
>>the human interocular, and the resolution isn't all that bad - perhaps she
>>meant that 15 feet is the cutoff for calculating distances within some
>>specified limit of precision?

>There is a theoretical limit to how precise they can get.  The size
>of a single digitized pixel might be enough to stop good distance
>calculations beyond 15.  A rough calculation, at 15 feet (~5 meters)
>with a stereo separation of 65mm is .74 degrees.  Depending
>of the angle of view and resolution of the camera....that just might
>not be very many pixels  (lets say 60 degree view with 1024 pixels
>17 pixels/degree or 12 pixels different..Now +/- 1 pixel could end
>up with almost half a meter (~1.5 feet) difference...pretty big...that
>is for something at 5 meters...what if it is at 20?

That's a compelling argument, but I'm not sure that the "for instance"
60 degree field of view is especially useful for this discussion. I
don't have access to technical data at present, but I believe it was mentioned
that the landscapes we've seen are mosaics of a pretty large number of
images, implying that the field of view of each image is fairly small.

Since the original post on this topic mentioned depth in the context of
building up a virtual reality model, it seems plausible that the accurate
construction of this model is the *main* reason for wanting accurate distance
measurements, in which case it is plausible that they might want the
readings to be accurate to much better than 1/2 meter - perhaps a few
centimeters?

Does anyone have access to the actual numbers for the lander stereo imager?

John R


------------------------------