STH-MD1 USER’S MANUAL ?2001 VIDERE DESIGN
12
problem for stereo processing, but does cause the display to look very dark
in low-light areas. You can add gamma correction to the displayed image
by choosing an appropriate gamma value in the slider under the right
display window (Figure 6-2).
The CMOS imagers have electronic exposure and gain controls to
compensate for varying lighting conditions. The exposure can vary from a
maximum of a full frame time to a minimum of one line time. Gain is an
additional amplification of the video signal, for low-light situations. It is
settable from 0 to 18 dB.
Both imagers are treated in exactly the same manner. It is not possible to
set a different exposure or gain on each imager.
Digitization control can operate in either manual or automatic mode.
Refer to Figure 6-1 for the controls in the video capture program. Manual
mode is the only currently supported mode for the STH-MD1/C.
In manual mode, the user program sets the exposure and gain. The
exposure and gain are based on a 0 to 100 scale. Here are some tips for
setting exposure and gain.
??In general, keep the gain as low as possible, since it introduces
additional noise into the system. Use it only if the exposure is set
to maximum, or if the exposure must be kept low to minimize
motion blur.
??Adjust the manual iris of the lens to as small an opening as
possible for your application, without having to use gain. This
will increase the depth of field and give better optical
performance.
6.4 Subsampling In many applications it is not necessary to work with the the full 1288 x
1032 pixel array. The CMOS imagers are capable of sampling the pixels
in the array. Sampling allows the video stream to send less data, for faster
frame rates or less bus activity. A sampled image shows the same scene as
the original image, but it uses fewer pixels to do so, and has less detail.
Sampling differs from subwindowing, which picks a rectangular portion of
the image, but doesn’t change its resolution.
The PC capture interface supports two types of subsampling. Decimation
is a sampling technique in which only a portion of the pixel values are sent
back to the host PC. Decimation takes place at the CMOS imagers, so
frame rates can be much higher. The STH-MD1/-C supports decimation
by 2 and by 4.
Binning is a subsampling technique in which several adjacent pixels are
averaged into one. Binning is superior to subsampling in that it reduces
video noise, sometimes quite substantially. However, binning does not cut
down on the video stream data rate for the STH-MD1/-C, since the binning
is done on the host PC.
Subsampling is always done in both the vertical and horizontal direction.
Subsampling x2 means that an image of size H x V will be transformed
into an image of size H/2 x V/2. Subsampling x4 transforms it to an image
of size H/4 x V/4.