14
14
3.3 Basic Camera Control Register-base Camera Control
The GEViCAM platform is register base. The command control is done with the 16-bit of address and
the 32-bit of data in hexadecimal. To send the “
Gain”
command of 255 (15dB) as an example, start
with the “
Write”
command “57” followed by the address “00 10” and the data “00 00 00 FF”. It receives
acknowledgement of “06” after the completion of command. The “
Read”
command is “52”.
3.3.1 Gain and Black Level Control
3.3.1.1 Gain Control (GainRaw):(0010)
The camera gain is controlled in the10-bit resolution from 6 dB to
42 dB (0~3FF). The variable gain amplifier is set by;
VGA Gain (dB) = (VGA code x 0.035 dB) + 5.3 dB
3.3.1.2 Black Level Control (BlackLevelRaw):(0011)
The video black level control is in the 8-bit resolution from 0 to 16
digital value out of 255 (0~FF). The factory default is set at 4.0
(digital value)
3.3.1.3 Dynamic Range and S/N Ratio
The GEViCAM platform is set the factory default at the best dy-
namic range. Since the platform performs at very low noise floor
providing overall signal to noise ratio of
60 dB
, the cameras are set to operate for the widest dynamic
range in the 12-bit resolution. Because of such a low noise floor, the images, which may look too dark
to human eyes, still contain very good signal once it is enhanced. On the contrary, if high gain is used,
the dark image may be visible but other brighter sections will be saturated and lost. Our
concept of a
digital imaging camera is to provide the best signal to noise ratio
and use full advantage of digiti-
zation for post image processing (can be real time).
However, those who need high gain operation, it has Max. 42 dB of internal gain adjustment, which is
extremely high gain!
The simplest measurement of S/N Ratio is using black level data deviation (standard-deviation) in a
large region of interest (ROI). The specific definition of SNR is 20 log (255/ rms-noise) and rms-noise
can be replaced by standard-deviation; SNR = 20 log (255/standard-deviation). 255 is saturation level
in 8-bit. Temporal noise, which is averaged in frame by frame data for the exact S/N measurement, is
ignored in this simple measurement. However, in the single frame, the large ROI deviation provides
good representation in practical value. In the 8-bit data, the popular software such as Photoshop can
measure the standard deviation.
3.3.2 Data Output Modes (0012)
3.3.2.1 12-bit, 10-bit, 8-bit Linear Output (PixelDepth): (00 00 00 00, - 01, -02)
The GEViCAM platform has a 12-bit high speed A/D converter and the output is selectable of 12, 10
and 8-bit (Factory Default: 8-bit). Both 10 and 8-bit conversions are done by taking the most significant
bits out of the 12 bit. All black and white cameras are set at the linear output.
3.3.2.2 8-bit LUT (Gamma) Output: (00 00 00 03)
The 12-bit to 8-bit conversion LUT (Look Up Table) is built in. The multiple knee Gamma (0.45) LUT is
embedded for the better dynamic range but limiting dark noise. All color cameras are set the factory
default to 8-bit and Gamma 0.45. Custom LUT is unloadable. See field upgrade in section 6.2.
3.3.2.3 16-bit Data Conversion: (00 00 00 20)
Many image process or analysis software uses 8-bit or 16-bit raw data input. This selection allows 12-
bit output to be used in 16-bit image input tool. The most significant 12-bits are used in 16-bit.
(Note: Italic description in parentheses is based on GigE Vision standard naming convention.)
GEViCAM Inc
Fig. 19
Содержание GP-151400
Страница 39: ...39 39 GEViCAM Inc MEMO ...
