51
A5 BINNING WITH CMOS SENSORS
A5
BINNING WITH CMOS SENSORS
Binning means the summation of single pixels to form larger pixels
and thereby improve the signal-to-noise ratio (SNR).
Binning: CCD image sensors
The term binning comes from scientific CCD image sensors. It is a
process of adding charges of two or more pixels before they are read
out of the sensor (pass the output amplifier) applied on very low light
signals. The prominent feature of charge-coupled-devices (CCD) is
the lossless transport or shifting of charge packages until an amplifier
circuit converts them into a voltage at the output, where the main
readout noise contribution occurs. Since the signal is increased
before it is read out, and the image sensor’s readout characteristics
remain unchanged, binning improves the SNR, but the image
sensor’s resolution is reduced. The lossless transport feature of CCD
image sensors makes binning possible.
Binning: CMOS image sensors
In general, binning is not possible in CMOS image sensors because
voltages are processed and no charges are transported. In each
pixel, the light generated charges are converted into voltages with the
readout noise contribution of these amplifiers. Therefore, as opposed
to CCD image sensors, if these voltages are combined, the readout
noise would also be combined, which would not have the same
positive effect on the SNR.
Nevertheless, such a summation or even an averaging would be
beneficial for the SNR, but with a smaller impact compared to CCD
image sensors. Since such CMOS binning cannot be done within the
image sensor, it either has to be done in the camera or in the
computer.
CMOS binning 1 – accumulation:
pixel values can accumulate,
causing an effective dynamic reduction or larger number formats,
because the result might exceed the original format -- two times
maximum 8 bit values will result in a 9 bit value. This is not a problem
as long as 12 bit values are accumulated and transported as 16 bit
images. However, in the case of scientific CMOS, if 16 bit values are
transmitted in 16 bit images, only two times 15 bit maximum values
are allowed. The signal of the binned pixels will be accumulated. Due
to the properties of readout noise; the increase of the noise itself will
only be as big as the square root of the number of binned pixels. The
SNR will improve and in addition, a reduction of the stored image
data is achieved.
CMOS binning 2 – averaging: pixel values can be averaged, which
has the same effect on the SNR as accumulation due to the
properties of noise. This would keep the image output format the
same and would reduce the amount of image data that can be
stored. When this type of CMOS binning is processed in the camera
it is called hardware binning. This should not be confused with real
binning in CCD image sensors, because the hardware that processes
this binning is not much different from the hardware in computer
processing. Therefore, the term hardware binning may be
misleading.
NOTE
The default binning
mode is
accumulation. Aver-
aging is not available
within Camware.
