The dynamic range of a charge-coupled device (CCD) or complementary metal oxide semiconductor (CMOS)
image sensor is typically specified as the maximum achievable signal
divided by the camera noise, where the signal strength is determined by
the full-well capacity, and noise is the sum of dark and read noises.
As the dynamic range of a device is increased, the ability to
quantitatively measure the dimmest intensities in an image (intrascene
performance) is improved. The inter scene dynamic range represents the
spectrum of intensities that can be accommodated when detector gain,
integration time, lens aperture, and other variables are adjusted for
differing fields of view.
Photodiode size determines, in part, the size of the depletion
wells--larger diodes having greater full-well capacity compared to
camera noise. Typical diode sizes in modern CCDs utilized in
photomicrography range from 4.5 to 24 microns with corresponding well
capacities of 20,000 to 600,000 electrons. Read noise is a combination
of all noise generated during readout of the device. This includes
noise from input clocking and fixed pattern, along with reset transistor
noise and amplifier output noise. Read noise is usually specified in
the performance data sheets that accompany a CCD sensor, with typical
values ranging from 10-20 electrons/pixel in high quality chips operated
at room temperature, and dropping to 2-5 electrons/pixel in
Peltier-cooled CCDs for scientific imaging applications. The dynamic
range is expressed in decibel units according to the following equation:
Dynamic Range = 20 x Log(Nsat/Nnoise)
where N(sat) is the linear full well capacity stated as the number of electrons and N(noise)
is the total value of the read and dark noise, also expressed as the
number of electrons. In a high-performance cooled CCD camera, the well
capacity is proportional to the size of the individual photodiode, such
that the maximum number of electrons stored is about 1000 times the
cross sectional area of each photodiode. Thus, a CCD with 6.7 x 6.7
micron photodiodes should have a maximum charge storage capacity (a
full-well capacity) of about 44,900 electrons (or holes). At a typical
readout rate of 1 MHz, the read noise for this CCD is about 10
electrons/pixel, which yields a dynamic range of 44,900/10 or 4,490. In
order to utilize the full range of grayscale levels available with this
dynamic range, the camera should have a 12-bit analog-to-digital (A/D)
converter capable of resolving 4096 gray levels. Controlling the size
of the read and dark noise is a critical factor in maintaining a high
dynamic range in these devices.
Higher performance cooled CCD sensors designed with low noise output
amplifiers and suitable for use in slow-scan imaging of photomicrographs
often have lower read noise and an extended dynamic range. As an
example, the Marconi Applied Technologies CCD39-01 sensor is a
back-illuminated, frame-transfer CCD having a square pixel size of 24
microns with a split output register allowing the utilization of quad
output amplifiers. The full well capacity of this device can reach a
level 300,000 electrons. Coupled with a readout noise root-mean-square (rms)
level of three electrons at 20 kilohertz (when cooled), the CCD39-01 is
capable of yielding a dynamic range of approximately 100,000:1.
To fully utilize the potential of this CCD, a 17-bit A/D converter
having 131,072 grayscale levels should be employed (although a 16-bit
A/D converter having 65,536 grayscale levels would also suffice).
The dynamic range of a particular CCD is dependent upon several
variables. Dark current is strongly influenced by temperature (Figure
1), doubling every 8 to 10 degrees Centigrade. At higher temperatures,
dark current is dominant, while at lower temperatures, dynamic range is
determined by the noise of the output amplifier. The amount of dark
charge collected in each pixel is dependent not only on the device
temperature, but also on the integration time and the storage time
before readout. The noise level is also proportional to the bandwidth
of the read-out amplifier, which is influenced by pixel transfer rate
and is thus affected by the clock frequency. As the clocking frequency
is increased, the number of dark current and shot noise electrons is
correspondingly decreased and less bandwidth is required by the output
amplifier and video-processing electronics. Integration time also
affects the dynamic range of a CCD, as illustrated in Figure 1. An
increase in the total integration time produces an increase in dark
current and a subsequent decrease in dynamic range, but this effect only
comes into play at integration times exceeding 5 minutes.
Bit depth refers to the binary range of possible grayscale values
utilized by the A/D converter to translate analog image information into
discrete digital values capable of being read and analyzed by a
computer. For example, the most popular 8-bit A/D converters have a
binary range of 2�(E8) or 256 possible values (Figure 2), while a 12-bit
converter has a range of 2�(E12) or 4,096 values, and a 16-bit
converter has 2�(E16), or 65,536 possible values. The bit depth of the
A/D converter determines the size of the gray scale increments, with
higher bit depths corresponding to a greater range of useful image
information available from the camera. Better results are obtained when
the signal is sampled at a level that is well beneath the limit
suggested by the readout noise. For instance, if the Marconi CCD39-01
is used with signal averaging, an 18-bit (262,144 grayscale levels) A/D
converter might be used to sample data at 1 part in 262,144. However,
the noise level statistics for this device indicate that image data
cannot be accurately measured to greater than 1 part in 100,000 without
signal averaging. Clearly, a 16-bit or 18-bit A/D converter will
produce better results when coupled to the Marconi CCD39-01 chip. In
contrast, Fujichrome Velvia, a fine-grained color transparency film, has
been demonstrated to produce less than 10-stops (1024 grayscale levels)
of dynamic range.
Table 1 presents the relationship between the number of bits used to
store digital information, the numerical equivalent in grayscale levels,
and the corresponding value in decibels (one bit equals approximately 6
dB). As illustrated in the table, if a 0.72 volt video signal were
digitized by an A/D converter with 1-bit accuracy, the signal would be
represented by two values, binary 0 or 1 with voltage values of 0 and
0.72 volts. Most digitizers found in digital cameras used in
photomicrography employ 8 bit A/D converters, which have 256 discrete
grayscale levels (between 0 and 255), to represent the voltage
amplitudes. A maximum signal of 0.72 volts would then be subdivided
into 256 steps, each step having a value of 2.9 millivolts.
Bit Depth and Dynamic Range of Charge-Coupled Devices
|
 |
|||||||||||||||||||||||||||||||||||||||||||||||||||||||
 |
 |
Table 1
The number of grayscale levels that must be generated in order to
achieve acceptable visual quality should be enough that the steps
between individual gray values are not discernible to the human eye.
The "just noticeable difference" in intensity of a gray-level image for
the average human eye is about two percent under ideal viewing
conditions. At most, the eye can distinguish about 50 discrete shades
of gray within the intensity range of a video monitor, suggesting that
the minimum dynamic range of an image should lie between 6 and 7 bits
(64 and 128 grayscale levels; Figure 2).
Digital images should have at least 8-bit resolution to avoid
producing visually obvious gray-level steps in the enhanced image when
contrast is increased during image processing. The effect of reducing
the number of grayscale levels on the appearance of photomicrographs can
be seen in Figure 3, which shows a black & white (originally 8-bit)
image of stained thin section of Solanum tuberosum
(potato) that is displayed at different resolutions ranging from 6-bit
(Figure 3(a)), down to 5-bit (Figure 3(b)), 4-bit (Figure 3(c)), and
3-bit (Figure 3(d)).
Improved digital cameras with CCDs capable of 12-bit resolution allow
investigators to display images with a greater latitude than is
possible with 8-bit images. This is possible because the appropriate
software can render the necessary shades of gray from a larger palette
(4,096 grayscale levels) for display on computer monitors, which
typically present images in 256 shades of gray. In contrast, an 8-bit
digital image is restricted to a palette of 256 grayscale levels that
were originally captured by the digital camera. As the magnification is
increased during image processing, the software can choose the most
accurate grayscales to reproduce portions of the enlarged image without
changing the original data. This is especially important when examining
shadowed areas where the depth of the 12-bit digital image allows the
investigator to visualize subtle details that would not be present in an
8-bit image.
The accuracy required for digital conversion of analog video signals
is dependent upon the difference between a digital gray-level step and
the rms noise in the camera output. CCD cameras with an internal A/D
converter produce a digital data stream that does not need to be
resampled and digitized in the computer. These cameras are capable of
producing digital data with up to 18-bit resolution (262,144 grayscales)
in high-end models, and are not constrained to the 0.72-volt signal
limitation of RS-170 video systems and utilize a wider analog voltage
range in their A/D converters. The major advantage of the large digital
range exhibited by CCD cameras lies in the signal-to-noise improvements
in the displayed 8-bit image and in the wide linear dynamic range over
which signals can be digitized.
Kenneth R. Spring - Scientific Consultant, Lusby, Maryland, 20657.
Michael W. Davidson - National High Magnetic Field Laboratory, 1800 East Paul Dirac Dr., The Florida State University, Tallahassee, Florida, 32310.