An set of example master dark frames taken with different fixed exposure settings and
acquired using the procedure just described with the AF51 camera is shown in figure 7.3.
Note the concentric and linear patterns as well as the gradation of bias signal across the
field. Note also how the pattern changes for different exposure settings.
For the relatively fast exposure of 1269 (top row images in figure 7.3) there is an
appearance of concentric rings that may be a reflection of the fixed pattern LENC gain
correction factor discussed in chapter 3. Note how this is greatly reduced when the longer
‘night mode’ multi-frame integration (on-chip) exposure is used (middle row in figure 7.3).
Note also that for the faster exposure time (top row) there are some completely black
pixels on the Y-only channel (left). These completely black pixels show no variation at all
in them – they all have the exact same value of ‘16’ without any variation despite being a
64-frame average. This indicates that the amount of charge on those pixels was so low
that it was below the ADC amplifier threshold so only the bias signal was output (for the
AF51, the bias signal – the lowest pixel value possible in the read out image – is 16 for the
Y channel of the YUYV camera stream). This occurred because at this relatively fast
exposure rate those pixels did not accumulate sufficient electrons by the type the rolling
shutter read out the pixel values in those pixels. Under these circumstances, no amount of
multi-frame averaging in software will be able to bring out any signal in those pixels at this
light level (remember that this is a dark field exposure, when the lens cap is removed we
can expect some light to get to those pixels and bias them sufficiently to overcome this
deficiency and read out some meaningful signal).
However, when a longer exposure time is allowed such as in the middle row pictures of
figure 7.3, sufficient charge has built up in all pixels across the field, even in complete
darkness due to thermal dark current signal accumulation, that they all read out signal
above the bias level. It is this property that makes the AF51 useful for imaging is low light
level situations.
The middle and lower panels of figure 7.3 shows a hyper-sensitive area which may be due
to small fluctuations in ADC amplifier performance and is one of the sources of dark field
variation that using a master dark for field correction is designed to compensate for in
quantitative work. The lower panel shows this area in greater detail and you can see,
comparing left (Y-only channel) and right (full YUYV-to_RGB conversion followed by
extraction of the intensity channel from an RGB-to-HSI transform) that we get slightly
sharper resolution with less vertical smearing using the full YUYV data to derive our
intensity information as opposed to using just the Y-values alone – but at the expense of
some increased noise fluctuations. Using more frames in the multi-frame averaging
process can reduce random noise however (I only used 64 frames for these pictures).
Colour streak artefact in low light conditions
In very low light levels with sufficiently fast exposure times such that pixels cannot read
above the ADC amplifier threshold, the AF51 camera shows a particular streaking artefact
where triangles appear to streak from bright pixels a bit like a tail behind a comet. This is
shown in the dark field images taken with the full YUYV colour signal from the camera in
figure 7.3 top, right.
OptArc AF51 Camera Page 92 of 99 User Guide v1.02
