Dalsa QuickCam, Руководство пользователя

Страница: 54 / 85

QuickCam User's Manual 52
03-032-10187-04 DALSA corp.
Analog Processing
Optimizing offset performance and gain in the analog domain allows you to
achieve a better signal-to-noise ratio and dynamic range than you would achieve
by trying to optimize the offset in the digital domain. As a result, perform all
analog adjustments prior to any digital adjustments.
1.
Analog gain is multiplied by the analog signal to increase the signal strength
before the A/D conversion. It is used to take advantage of the full dynamic
range of the A/D converter. For example, in a low light situation the
brightest part of the image may be consistently coming in at only 50% of the
DN. An analog gain of 6 dB (2x) will ensure full use of the dynamic range of
the A/D converter. Of course the noise is also increased.
2. The analog offset or black level is an "artificial" offset introduced into the
video path to ensure that the A/D is functioning properly. The analog offset
should be set so that it is at least 3 times the rms noise value at the current
gain.
Digital Processing
To optimize camera performance, digital signal processing should be completed
after any analog adjustments.
1.
Fixed pattern noise (FPN) calibration is used to subtract away individual
pixel dark current.
2. The digital offset enables the subtraction of the "artificial" A/D offset (the
analog offset) so that application of the PRNU coefficient doesn’t result in
artifacts at low light levels due to the offset value.
3. Photo-Response Non-Uniformity (PRNU) coefficients are used to correct the
difference in responsivity of individual pixels (i.e. given the same amount of
light different pixels will charge up at different rates) and the change in light
intensity across the image either because of the light source or due to optical
aberrations (e.g. there many be more light in the center of the image). PRNU
coefficients are multipliers and are defined to be of a value greater than or
equal to 1. This ensures that all pixels will saturate together. When using
PRNU correction, it is important that the A/D offset and Fixed Pattern Noise
(FPN) or per pixel offsets are subtracted prior to the multiplication by the
PRNU coefficient. The subtraction of these 2 components ensure that the
video supplied to the PRNU multiplier is nominally zero and zero
multiplied by anything is still zero resulting in no PRNU coefficient induced
FPN. If the offset is not subtracted from the video then there will be artifacts
in the video at low light caused by the multiplication of the offset value by
the PRNU coefficients.
4. Background subtract and digital gain are used to increase image contrast
after FPN and PRNU calibration. It is useful for systems that process 8-bit
data but want to take advantage of the camera’s 12-bit digital processing
chain. For example, if you find that your image is consistently between 128
and 255DN(8-bit), you can subtract off 128 and then multiply by 2 (by setting
the digital gain to 8192) to get an output range from 0 to 255.