dynamic - dynamic range
1 dynamic range of a CCD image nsor
The dynamic range (dyn CCD ) is defined as the ratio of the maximum possible signal (full well capacity), ver-sus the total noi signal (in the dark). The data is ex-presd in decibels [dB] or is dimensionless:
dyn readout noi e fullwell capacity e CCD =
--66@@
dyn 20log readout noi e fullwell capacity e dB CCD =$--c m 666@
@
@
The dynamic or dynamic range of a CCD-camera system is a widely ud term to characterize the ability of a camera system to measure or image light/dark differences. In the field of photography, the dynamic range is analogous to the contrast range. However, each manufacturer defines dynamic range differently. A distinction must be made, between the “dynamic range of a CCD image nsor”, “dynamic range of an analog-to-digital-conversion” and “usable dynamic range”.
an analog-to-digital (A/D) converter. Therefore, the light signals (photons) are converted into digital values. The analog-to-digital converters have a given resolution or dynamic range, that in most cas is given as a power of ba 2, (2x). This means that an 8 bit resolution corresponds to 256 steps, which can be ud to subdivide or convert the full scale voltage signal. CCD camera manufacturers usually optimize a combination of the dynamic range of the CCD image nsor, gain and conversion factor (defined as average conversion ratio that takes x electrons to generate one count in the image) to optimize the dynamic range of the CCD image nsor by the analog-to-digital converter.
The dynamic range of the digitization is not identical
to the usable dynamic range.
parameter
Sony ICX285AL Kodak KAI-1020Kodak KAI-11000fullwell capacity [e -]18 00040 00060 000readout noi [e -]
5
911dynamic range 3 600 : 1 4 444 : 1 5 454 : 1dynamic range [dB]
71.1
72.9
74.7
table 1: common CCD image nsor specifications
Upon examination of the specifications, the ICX285 could be digitized using 4096 steps (corresponds to {212} 12 bit resolution), while the KAI-11000 should be digitized using 8192 steps (c
orresponds to 13 bit resolution). This references to maximum values which are explained in detail in the usable dynamic range ction.
Furthermore, additional gain makes n only if the dynamic range of the A/D-converter is smaller than the dynamic range of the CCD image nsor. For example, if the dynamic range of the CCD is 70 dB, and the dynamic range of the A/D-converter is 48 dB, a uful gain or amplification would be 22 dB. Alternatively, a programmable gain would increa the performance of such a system.
The resolutions above directly correspond to the
theoretical maximum limit of the converter devices. Analog-to-digital converters have an average con-version uncertainty of 0.4 - 0.7 bit, which reduces the resolution for practical applications by 1 bit. If the camera system is not limited in its dynamic range by A/D converter discrepancies, it is uful to inflate the A/D converter resolution by 1 or 2 bits. This is achieved by electronically adding a minor amount to the signal a so-called offt signal, so that the lower limit is solely provided via the image nsor and read out amplifier noi, with some resolution sacrifices from the converter.
resolution [bit]x => 2x dynamic range A/D conversion [digitizing steps]dynamic range A/D conversion [dB]825648.210 1 02460.212 4 09672.31416 38484.316
65 536
96.3
table 2: dynamic range of binary resolution
2 digitization or A/D conversion dynamic range
Generated charge carriers are usually converted into voltage signals through an optimized read out circuit. The signals are amplified and finally digitized by
dynamic - dynamic range
4 comparing 12 bit and 8 bit dynamic range
This low light application example was recorded using a 12 bit cooled CCD camera system. The full scale, 12 bit image is completely dark. However, when the image is scaled to the 54-76 counts range, the image becomes visible. Although noi can be en due to a weak light signal, the image can be clearly en. Under similar conditions, but recorded with an 8 bit system
3 usable dynamic range
The actual usable dynamic range of a camera system is determined by the general ttings or adjustments of the system (e.g., the quality of the dynamic range of the CCD nsor, and the compatibility between the read out circuit’s dynamic range or digitization). For example, if a full well capacity of 18 000 electrons with a CCD image nsor dynamic range ratio of 4500:1 is given, getting optimal range with a 12 bit A/D converter (4096 gray levels) is possible with a conversion factor of 4.4 electrons per step (count ). For very low noi cameras, it is sometimes advantageous to shift the signal resolution towards the lower noi limit and ignore the total dynamic range of the CCD image nsor, particularly if it is of minor interest in the speci-fic application.
A particular application can also influence the usable dynamic range. Although some manufacturers attempt to define their dynamic ranges as being independent of application, the applications have a direct effect on the limits of the dynamic range. For example, in medium to high light level applications, which are photon noi limited, the intrascene dynamics may be
smaller than the system’s dynamic range.
Figure 1
Weakly illuminated image with 12bit total dynamic displayed with
255 gray levels. scaled: 0-4095 -> 0-255 graph shows readout line
Figure 2
Same image as above with 12bit total dynamic displayed with 255
gray levels. scaled: 53-70 -> 0-255 graph shows readout line
Figure 3
Same image as above but with 8bit total dynamic displayed with 255 gray levels. scaled: 7-9 -> 0-255 graph shows readout line
europe
PCO AG
Donaupark 11
93309 Kelheim, Germany fon +49 (0)9441 2005 50fax +49 (0)9441 2005 20info@pco.de www.pco.de
america
The Cooke Corporation 6930 Metroplex Drive
Romulus, Michigan 48174, USA tel (248) 276 8820fax (248) 276
and scaling, the image can only be procesd by the human eye, using the unique capabilities of the human brain, and could not be image procesd in a computer.
